Purpose

Guide product managers through evaluating whether to scale, test, or kill an acquisition channel based on unit economics (CAC, LTV, payback), customer quality (retention, NRR), and scalability (magic number, volume potential). Use this to make data-driven go-to-market decisions and optimize channel mix for sustainable growth.

This is not a channel strategy framework—it's a financial lens for channel evaluation that helps you avoid scaling unprofitable channels or killing channels with fixable problems. Use when deciding how to allocate marketing budget across channels.

Input

Works best with: The acquisition channel you're evaluating (e.g., paid search, outbound SDR, partner referrals). Also useful: Any metrics you already have — CAC, LTV, payback period, retention/NRR by channel — plus company stage and the decision on the table (scale, test, or kill).

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The advisor opens by asking which channel you're evaluating and what data you have.

Example invocation: Evaluate our paid LinkedIn channel: CAC $2,400, LTV $9,000, 14-month payback, flat retention vs. organic.

Key Concepts

The Channel Evaluation Framework

A systematic approach to evaluate acquisition channels:

1. Unit Economics — What does it cost to acquire, and what's the return?

   • CAC (Customer Acquisition Cost)
   • LTV (Lifetime Value)
   • LTV:CAC ratio
   • Payback period

2. Customer Quality — Do customers from this channel stick around and expand?

   • Cohort retention rate (by channel)
   • Churn rate (by channel)
   • NRR (Net Revenue Retention by channel)
   • Expansion rate

3. Scalability — Can this channel sustain growth at the volume you need?

   • Magic Number (S&M efficiency)
   • Addressable volume (TAM of channel)
   • Saturation risk (diminishing returns)
   • CAC trend (increasing, stable, decreasing)

4. Strategic Fit — Does this channel align with your go-to-market strategy?

   • Customer segment match (SMB vs. enterprise)
   • Sales motion compatibility (PLG vs. sales-led)
   • Brand positioning alignment

Decision Matrix

Anti-Patterns (What This Is NOT)

• Not vanity metrics: "We got 10,000 signups!" means nothing if they churn in 30 days
• Not CAC-only thinking: Low CAC with terrible retention is worse than high CAC with great retention
• Not ignoring payback: 5:1 LTV:CAC with 36-month payback is a cash trap
• Not scaling broken channels: Pouring money into inefficient channels accelerates failure

When to Use This Framework

Use this when:

• Evaluating whether to scale a new channel (content, paid, events, etc.)
• Deciding how to allocate marketing budget across channels
• Assessing whether to kill an underperforming channel
• Comparing channels to optimize ROI
• Planning annual marketing budget allocation

Don't use this when:

• Channel is brand-new (<3 months, <100 customers) — not enough data
• You're testing channel fit (this is for evaluation, not experimentation)
• Strategic channels (e.g., enterprises require field sales regardless of CAC)
• You don't have channel-level data (need to track CAC, retention by source)

Facilitation Source of Truth

Use workshop-facilitation as the default interaction protocol for this skill.

It defines:

• session heads-up + entry mode (Guided, Context dump, Best guess)
• one-question turns with plain-language prompts
• progress labels (for example, Context Qx/8 and Scoring Qx/5)
• interruption handling and pause/resume behavior
• numbered recommendations at decision points
• quick-select numbered response options for regular questions (include Other (specify) when useful)

This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.

Application

This interactive skill asks up to 4 adaptive questions, offering 3-5 enumerated options at decision points.

Step 0: Gather Context

Agent asks:

"Let's evaluate this acquisition channel. Please provide:

Channel details:

• Channel name (e.g., Google Ads, content marketing, outbound sales, partnerships)
• How long have you been using this channel? (months)
• Current monthly spend on this channel

Customer acquisition:

• Customers acquired per month (from this channel)
• CAC for this channel (if known, otherwise we'll calculate)

Business context:

• Blended CAC (across all channels)
• Blended LTV
• Current MRR/ARR
• Target growth rate (% MoM or YoY)

You can provide estimates if you don't have exact numbers."

Step 1: Evaluate Unit Economics

Agent calculates (if not provided):

CAC = Monthly Spend / Customers Acquired per Month

Agent asks:

"Now let's compare this channel's unit economics to your blended metrics.

Channel Unit Economics:

• Channel CAC: $___
• Blended CAC (all channels): $___
• Channel LTV: $___ (if known; otherwise we'll use blended LTV as proxy)
• Blended LTV: $___

Questions:

1. Do customers from this channel have similar LTV to other channels?

   • Similar (use blended LTV)
   • Higher (they upgrade more, stick around longer)
   • Lower (they churn faster or are smaller deals)
   • Unknown (need to analyze cohort data)

2. What's the payback period for this channel?

   • We can calculate: CAC / (Monthly ARPU × Gross Margin %)
   • Or you can provide it"

Based on answers, agent calculates:

• LTV:CAC ratio for channel
• Payback period
• Comparison to blended metrics

Agent flags:

• ✅ If LTV:CAC >3:1 and payback <12 months: "Strong unit economics"
• ⚠️ If LTV:CAC 2-3:1 or payback 12-18 months: "Marginal unit economics"
• 🚨 If LTV:CAC <2:1 or payback >18 months: "Poor unit economics"

Step 2: Assess Customer Quality

Agent asks:

"How do customers from this channel perform compared to other channels?

Retention & Expansion:

1. What's the churn rate for customers from this channel?

   • Lower than blended (they stick around longer)
   • Same as blended (no difference)
   • Higher than blended (they churn faster)
   • Unknown (need cohort analysis)

2. What's the NRR for customers from this channel?

   • Higher than blended (they expand more)
   • Same as blended (no difference)
   • Lower than blended (they contract or churn more)
   • Unknown (need cohort analysis)

3. What's the customer profile from this channel?

   • Ideal customer profile (ICP) — perfect fit
   • Close to ICP — mostly good fit
   • Off ICP — many poor-fit customers
   • Unknown"

Based on answers, agent evaluates:

• ✅ High quality: Lower churn, higher NRR, ICP match
• ⚠️ Medium quality: Similar to blended, mostly good fit
• 🚨 Low quality: Higher churn, lower NRR, off ICP

Agent flags:

• If high quality: "Premium channel—customers are better than average"
• If low quality: "Quality problem—customers aren't sticking or expanding"

Step 3: Evaluate Scalability

Agent asks:

"Can this channel scale to meet your growth targets?

Efficiency & Volume:

1. What's the S&M efficiency for this channel (Magic Number)?

   • Calculate: (New MRR from channel × 4) / Channel S&M Spend
   • Or provide if known

2. What's the addressable volume for this channel?

   • Large (can scale 10x+ from current spend)
   • Medium (can scale 2-5x)
   • Small (near saturation, maybe 1.5x)
   • Unknown

3. What's the CAC trend for this channel?

   • Decreasing (getting more efficient over time)
   • Stable (consistent CAC)
   • Increasing (diminishing returns, saturation)
   • Unknown (too early to tell)

4. How much growth do you need from acquisition?

   • We'll calculate: Target growth - expansion/retention growth = acquisition gap"

Based on answers, agent evaluates:

• ✅ Highly scalable: Magic number >0.75, large volume, stable/decreasing CAC
• ⚠️ Moderately scalable: Magic number 0.5-0.75, medium volume, stable CAC
• 🚨 Not scalable: Magic number <0.5, small volume, increasing CAC

Step 4: Deliver Recommendations

Agent synthesizes:

• Unit economics (LTV:CAC, payback)
• Customer quality (retention, NRR, ICP fit)
• Scalability (magic number, volume, CAC trend)
• Strategic fit

Agent offers 3-4 recommendations:

Recommendation Pattern 1: Scale Aggressively

When:

• LTV:CAC >3:1 AND
• Payback <12 months AND
• Customer quality good or better AND
• Magic Number >0.75 AND
• Addressable volume large

Recommendation:

"Scale this channel aggressively — Excellent economics + scalability

Unit Economics:

• CAC: $___
• LTV: $___
• LTV:CAC: ___:1 ✅ (>3:1 threshold)
• Payback: ___ months ✅ (<12 months)

Customer Quality:

• Retention: [Better than / Same as / Worse than] blended
• NRR: [Higher / Same / Lower]
• ICP Fit: [High / Medium / Low]

Scalability:

• Magic Number: ___ ✅ (>0.75 = efficient)
• Addressable Volume: Large
• CAC Trend: [Stable / Decreasing]

Why this is a winner:

• Every $1 spent returns $__ in LTV
• Payback in under a year = fast cash recovery
• [Customer quality insight]
• Can scale 5-10x from current spend

How to scale:

1. Increase budget by 50-100% next month
   • Current: $___ /month → Target: $___ /month
2. Monitor key metrics weekly:
   • CAC (should stay <$___)
   • Magic Number (should stay >0.75)
   • Customer quality (retention, NRR)
3. Scale until:
   • CAC increases >20% (saturation signal)
   • Magic Number drops <0.75 (efficiency declining)
   • Volume caps out

Expected impact:

• Current: ___ customers/month
• Target (2x spend): ___ customers/month
• MRR impact: +$___/month
• Payback: Still ~___ months even at 2x scale

Risk: Low. Strong unit economics support aggressive scaling."

Recommendation Pattern 2: Test & Optimize

When:

• LTV:CAC 2-3:1 OR
• Payback 12-18 months OR
• Customer quality average OR
• Magic Number 0.5-0.75

Recommendation:

"Test & optimize before scaling — Marginal economics, fixable

Current State:

• CAC: $___
• LTV: $___
• LTV:CAC: ___:1 ⚠️ (2-3:1 = marginal)
• Payback: ___ months ⚠️ (12-18 months)
• Magic Number: ___ ⚠️ (0.5-0.75 = acceptable, not great)

Customer Quality:

• Retention: [Same as blended / Slightly worse]
• NRR: [Same / Lower]
• Issue: [Specific problem, e.g., "Higher churn in first 90 days"]

Diagnosis: [One of these:]

• High CAC: Spending too much to acquire
• Low LTV: Customers churn too fast or don't expand
• Poor targeting: Attracting off-ICP customers
• Inefficient conversion: High cost-per-click but low conversion rate

How to optimize:

If CAC is the problem:

1. Improve conversion rate (optimize landing pages, offer, onboarding)
2. Reduce cost-per-click (better targeting, ad creative)
3. Shorten sales cycle (faster qualification, better demos)

If LTV is the problem:

1. Improve onboarding for customers from this channel
2. Target higher-value segments within channel
3. Add expansion plays (upsell, cross-sell)

If targeting is the problem:

1. Narrow audience (exclude poor-fit segments)
2. Improve messaging (attract better-fit customers)
3. Add qualification step (reduce poor-fit signups)

Timeline:

• Spend 4-8 weeks optimizing
• Track CAC and LTV weekly
• Target: LTV:CAC >3:1, payback <12 months
• If you hit targets: scale
• If you can't fix it: consider killing

Don't scale yet: Current economics are break-even at best. Fix first, then scale."

Recommendation Pattern 3: Kill or Pause

When:

• LTV:CAC <1.5:1 AND
• No clear path to improvement

Recommendation:

"Kill this channel (or pause) — Economics don't support investment

Why:

• CAC: $___
• LTV: $___
• LTV:CAC: ___:1 🚨 (<2:1 = unsustainable)
• Payback: ___ months 🚨 (>18 months = cash trap)

Problem:

• You're spending $___ to acquire a customer worth $___
• [Losing money / Barely breaking even / Taking too long to recover cost]

Customer Quality:

• Retention: [Worse than blended]
• NRR: [Lower]
• ICP Fit: [Poor]

What's broken: [Specific diagnosis:]

• CAC too high (spending $___ vs. blended $___)
• LTV too low (customers churn at ___% vs. blended ___%)
• Both (bad unit economics from both sides)

Should you fix or kill?

Fix if:

• You have a hypothesis to improve CAC by 50%+ (better targeting, conversion)
• You have a hypothesis to improve LTV by 50%+ (better onboarding, ICP focus)
• This is a strategically important channel (e.g., enterprise requires field sales)

Kill if:

• No clear path to 3:1 LTV:CAC
• Better channels available (reallocate budget there)
• Small addressable volume (not worth fixing)

Recommendation: Kill and reallocate budget

Reallocate to:

• Channel X (LTV:CAC = ___:1, can scale)
• Channel Y (Magic Number = ___, efficient)

What to do with budget:

• Current channel spend: $___/month
• Reallocate to [top-performing channel]
• Expected impact: [better CAC, better LTV, faster payback]

Exception: If this channel is <10% of total S&M spend, just pause it. Not worth fixing."

Recommendation Pattern 4: Invest to Learn (Strategic Channel)

When:

• Poor unit economics BUT
• Strategic importance (enterprise channel, brand building, long-term)

Recommendation:

"Continue, but cap investment — Strategic value > short-term ROI

Financial Reality:

• CAC: $___
• LTV: $___
• LTV:CAC: ___:1 (below 3:1 threshold)
• Payback: ___ months (long)

Why continue despite poor economics:

• [Strategic reason: e.g., "Enterprise segment requires field events, but deals are 12-month sales cycles"]
• [Brand building: e.g., "Conferences build brand awareness that drives inbound long-term"]
• [Market positioning: e.g., "Need to be present in this channel for credibility"]

How to manage:

1. Cap spend — Don't scale until economics improve
   • Current: $___/month
   • Cap at: $___/month (hold steady)
2. Track leading indicators — Don't just look at short-term CAC/LTV
   • Pipeline influence
   • Brand awareness lift
   • Referral rate from this channel
3. Re-evaluate quarterly
   • If economics improve (LTV:CAC >3:1): scale
   • If economics stay poor: reconsider strategy

Timeline:

• Give it [6-12 months] to show results
• If no improvement: kill or reduce drastically

Risk: You're subsidizing growth. Make sure it's worth it."

Step 5: Compare Across Channels (Optional)

If user has multiple channels, agent can generate:

Budget allocation recommendation:

1. Scale: Content (highest efficiency)
2. Scale: Google Ads (strong economics)
3. Optimize: Outbound (improve magic number)
4. Kill: Events (reallocate budget)

Examples

See examples/ folder for sample conversation flows. Mini examples below:

Example 1: Scale (Content Marketing)

Channel: Organic content (blog, SEO)

• CAC: $200
• LTV: $3,000
• LTV:CAC: 15:1
• Payback: 3 months
• Magic Number: 1.8
• Customer quality: High (lower churn, higher NRR)

Recommendation: Scale aggressively. Exceptional unit economics, fast payback, high-quality customers. Increase content spend 2-3x.

Example 2: Optimize (Paid Search)

Channel: Google Ads

• CAC: $800
• LTV: $2,000
• LTV:CAC: 2.5:1
• Payback: 14 months
• Magic Number: 0.6
• Customer quality: Lower (higher churn in first 90 days)

Recommendation: Test & optimize before scaling. CAC is high, onboarding is weak for this segment. Improve landing page, target higher-intent keywords, better onboarding for paid customers.

Example 3: Kill (Trade Shows)

Channel: Industry events

• CAC: $20,000
• LTV: $30,000
• LTV:CAC: 1.5:1
• Payback: 30 months
• Magic Number: 0.2
• Customer quality: Low (off-ICP, many tire-kickers)

Recommendation: Kill. CAC too high, payback too long, poor customer quality. Reallocate budget to content and paid search.

Common Pitfalls

Pitfall 1: Scaling Broken Channels

Symptom: "Let's 10x our Google Ads spend!" (LTV:CAC is 1.5:1)

Consequence: You accelerate cash burn without improving unit economics. Lose money faster.

Fix: Only scale channels with LTV:CAC >3:1 and payback <12 months. Fix broken channels before scaling.

Pitfall 2: Ignoring Customer Quality

Symptom: "CAC is only $100!" (but customers churn in 30 days)

Consequence: Low CAC means nothing if LTV is also low. You're acquiring churners, not customers.

Fix: Track cohort retention and NRR by channel. Low CAC + high churn = bad channel.

Pitfall 3: Celebrating Vanity Metrics

Symptom: "We got 10,000 signups from this campaign!" (5% convert to paid)

Consequence: Signups don't pay bills. CAC is calculated on paid customers, not signups.

Fix: Track CAC on paid customers only. Ignore vanity metrics like signups, impressions, clicks.

Pitfall 4: Averaging Across Channels

Symptom: "Blended CAC is $500" (but hiding that one channel is $10K CAC)

Consequence: Bad channels hide in blended metrics. You don't know which channels to kill.

Fix: Track CAC, LTV, payback by channel. Compare channels individually.

Pitfall 5: Short-Term CAC Optimization

Symptom: "We reduced CAC 50%!" (by targeting low-intent, low-LTV customers)

Consequence: CAC dropped but so did LTV. Unit economics got worse, not better.

Fix: Optimize for LTV:CAC ratio, not CAC alone. Higher CAC with higher LTV is better.

Pitfall 6: Ignoring Payback Period

Symptom: "LTV:CAC is 6:1, this channel is amazing!" (payback is 48 months)

Consequence: You run out of cash before recovering CAC. Great ratio, terrible cash flow.

Fix: Pair LTV:CAC with payback period. 3:1 with 8-month payback beats 6:1 with 36-month payback.

Pitfall 7: Killing Channels Too Early

Symptom: "This channel didn't work after 2 weeks"

Consequence: Channels need time to optimize. Killing too early wastes learning.

Fix: Give channels 3-6 months and 100+ customers before evaluating. Track trends, not snapshots.

Pitfall 8: Over-Relying on One Channel

Symptom: "90% of our customers come from Google Ads"

Consequence: Algorithm change, competitor outbids you, channel saturates = business grinds to halt.

Fix: Diversify channels. No single channel should be >50% of new customer acquisition.

Pitfall 9: Forgetting Incrementality

Symptom: "This retargeting campaign has great ROI!" (but customers would've converted anyway)

Consequence: You're paying for conversions that would happen organically. Inflated ROI.

Fix: Test incrementality with holdout groups. Only count truly incremental conversions.

Pitfall 10: Strategic Channels Without Limits

Symptom: "Enterprise events are strategic, we can't stop!" (losing $500K/year)

Consequence: "Strategic" becomes an excuse for burning cash indefinitely.

Fix: Cap spend on strategic channels. Set timeline for improvement (6-12 months). If no progress, kill.

References

Related Skills

• saas-economics-efficiency-metrics — CAC, LTV, payback, magic number calculations
• saas-revenue-growth-metrics — NRR, churn, cohort analysis by channel
• finance-metrics-quickref — Fast lookup for channel evaluation metrics
• feature-investment-advisor — Similar ROI framework for feature decisions
• business-health-diagnostic — Broader business health assessment

External Frameworks

• Brian Balfour (Reforge): Channel-product fit framework
• David Skok: "SaaS Metrics" — CAC, LTV, and payback for channels
• Tomasz Tunguz: SaaS channel benchmarking
• First Round Review: "How to Find and Scale Your Growth Channels"

Provenance

• Adapted from research/finance/Finance_For_PMs.Putting_It_Together_Synthesis.md (Decision Framework #2)
• Channel economics from research/finance/Finance for Product Managers.md

Purpose

Guide product managers through designing multi-agent workflows—breaking complex, repetitive PM tasks into parallel, specialized AI agents rather than linear, sequential processes or manual execution. Use this to transition from "document-heavy administrator" to "systems-level orchestrator" who coordinates a "living system" of AI agents, human teams, and market data interacting continuously.

Key Shift: From linear project management (one task at a time) to orchestration (multiple agents working simultaneously, each with clear boundaries and handoffs).

This is not about prompt writing—it's about architecting workflows where AI agents handle repetitive research, synthesis, and validation while PMs focus on strategy and decision-making.

Input

Works best with: The workflow or recurring task you want to orchestrate — described in a sentence or two, however manual or messy it is today. Also useful: Where it breaks down now (too slow, too sequential, too dependent on you), the tools your team already uses, and whether you've worked through context-engineering-advisor first (it's the prerequisite discipline).

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The advisor opens by asking which PM workflow eats the most of your week, then walks the four orchestration dimensions against it.

Example invocation: Design an orchestration for our weekly competitive intel: today one PM spends 6 hours scraping, summarizing, and briefing — sequentially.

Key Concepts

Orchestration vs. Project Management

Critical Insight: Orchestration is not about replacing humans—it's about force-multiplying human judgment by automating repetitive, time-consuming tasks.

The Four Dimensions of Orchestration

1. Coordination of Multi-Agent Workflows

Breaking complex tasks into specialized agents that run in parallel.

Example:

• Manual (Old): PM spends 8 hours compiling competitive intel, then 4 hours synthesizing customer feedback, then 3 hours identifying roadmap gaps = 15 hours sequentially
• Orchestrated (New): Three agents run simultaneously:
  • Agent A: Competitive intel (research agent)
  • Agent B: Customer synthesis (synthesis agent)
  • Agent C: Roadmap gap analysis (analysis agent)
  • Total time: 8 hours (limited by slowest agent), PM reviews outputs in 2 hours = 10 hours total, 5 hours saved

Key Principle: Shift from manual selection to hypothesis orchestration—agents generate hypotheses, PM validates and decides.

2. Leadership of Cross-Functional AI Pods

Governing diverse teams (data scientists, ML engineers, compliance, ethicists) to ensure solutions are scalable, ethical, and aligned.

What it includes:

• Embedding diversity-aware workflows
• Risk management (not afterthought)
• Ethical orchestration (ensuring AI doesn't "go rogue")
• Cross-functional alignment (engineering, compliance, design)

PM Role: Guardian of Governance—ensures AI systems reflect company values.

3. Launch Control Tower Function

Real-time monitoring of organizational readiness across functions using agentic systems to flag gaps before critical failures.

What it monitors:

• Support readiness (docs, training, escalation paths)
• Marketing readiness (messaging, assets, GTM plan)
• Operations readiness (infrastructure, scaling, monitoring)

Key Principle: Agentic systems act as early warning system—flag gaps before they become blockers.

4. Strategic Intent Alignment (Context Engineering Applied)

Feeding AI agents the correct mix of mission, constraints, and priorities to ensure automated decisions reflect company values.

Connection: This is context engineering at the orchestration layer. See context-engineering-advisor for foundations.

What agents need:

• Product constraints (what we will/won't build)
• Strategic priorities (what matters most right now)
• Operational definitions (shared glossary)
• Evidence standards (what counts as validation)

The Four AI Management Workflows (Productside Blueprint)

Every PM must master these workflows to move fast while staying grounded:

1. Context Engineering ✅ (Foundation)

   • Create AI workspace that remembers product domain, research, JTBD, personas, constraints
   • Skill: context-engineering-advisor

2. Synthetic Evals 📋 (Quality Assurance)

   • Automated validation tests for AI reasoning
   • Generate synthetic data, run workflows against traces
   • Eliminates 80% of hallucination risk

3. Agentic Workflows ← We're here

   • Agents handle repetitive tasks (competitive intel, customer synthesis, roadmap gaps)
   • PM focuses on strategy

4. Vibe Coding 📋 (Rapid Prototyping)

   • Generate clickable prototypes from context workspace
   • Collapse feedback loops from weeks to hours
   • Connection: pol-probe-advisor (Vibe-Coded PoL Probes)

AI-Shaped Problems (Teresa Torres)

What makes a problem "AI-shaped"?

• Previously difficult to scale due to human involvement (e.g., synthesizing 50 user interviews)
• Falls short with current non-AI solutions (e.g., manual competitive tracking)
• Requires consistency at scale (e.g., risk analysis across 100 features)

Key Insight: "While AI makes building easier, choosing what to build remains the primary challenge." Orchestration helps with the "building" part so PMs can focus on "choosing."

The Four Big Risks (Marty Cagan, AI Era)

The orchestrator manages these risks across the organization:

Anti-Patterns (What This Is NOT)

• Not about replacing PMs: Orchestration amplifies judgment, doesn't eliminate it
• Not about automating everything: Some tasks require human empathy and context
• Not about complexity for its own sake: Only orchestrate when it saves significant time or improves quality
• Not about "set it and forget it": Orchestrated workflows require monitoring and maintenance

When to Use This Skill

✅ Use this when:

• You have repetitive PM tasks that take 5+ hours per week (competitive analysis, customer synthesis, roadmap maintenance)
• You're doing sequential work that could be parallelized (research, then synthesis, then analysis)
• You need consistency at scale (analyze 50 features for risk, synthesize 100 customer interviews)
• You're spending time on execution instead of strategy

❌ Don't use this when:

• The task is one-time or infrequent (not worth orchestration overhead)
• Human judgment is critical at every step (empathy-driven work)
• The task is already fast enough (don't over-engineer)
• You haven't built context engineering foundations first (see context-engineering-advisor)

Application

This interactive skill uses adaptive questioning to design multi-agent workflows step-by-step.

Step 0: Gather Context

Agent asks:

Before we design your orchestrated workflow, let's gather context:

Current PM Work:

• What repetitive tasks take the most time each week? (competitive analysis, customer synthesis, roadmap updates, etc.)
• How much time do you spend on these tasks? (hours per week)
• Are these tasks sequential (one after another) or could they run in parallel?

Desired Outcome:

• What would "good enough" automation look like? (80% reduction in manual work? Higher consistency?)
• What parts require human judgment? (final decisions, empathy, ethical considerations)

Technical Context:

• Do you have context engineering foundations? (constraints registry, operational glossary—see context-engineering-advisor)
• What AI tools do you currently use? (Claude Projects, ChatGPT, custom agents)

You can describe briefly or paste examples.

Step 1: Assess if Task is AI-Shaped

Agent asks:

Let's determine if your task is suited for orchestration. Which statement best describes your situation?

Select one:

1. "I spend 10+ hours per week on repetitive research/synthesis tasks"

   • Examples: Compiling competitive intelligence, synthesizing customer feedback, tracking feature requests
   • Assessment: High potential for orchestration

2. "I have a task that requires multiple sequential steps, and each step is time-consuming"

   • Examples: Research competitors → Synthesize findings → Identify gaps → Update roadmap (15 hours total, done sequentially)
   • Assessment: Good candidate for parallelization

3. "I need consistency at scale—analyzing many similar items"

   • Examples: Risk analysis for 50 features, synthesizing 100 user interviews, tracking 20 competitors
   • Assessment: Perfect for orchestration (agents ensure consistency)

4. "I have a task that's mostly human judgment and empathy"

   • Examples: Stakeholder negotiation, team coaching, ethical trade-offs
   • Assessment: Not a good candidate for orchestration (keep human-driven)

User response: [Selection]

Agent analyzes:

Based on your selection:

• Options 1-3: Proceed to workflow design
• Option 4: Recommend keeping task human-driven; suggest alternative approaches (e.g., use AI for research, but keep decision-making human)

Step 2: Decompose Task into Agent-Specific Subtasks

Agent asks:

Let's break your task into discrete subtasks, each handled by a specialized agent.

For the task you selected, what are the distinct steps?

Example Decomposition (Competitive Intelligence):

Manual Process (Old):

1. Research 10 competitors' product pages (3 hours)
2. Track recent feature launches (2 hours)
3. Analyze pricing changes (2 hours)
4. Synthesize into competitive landscape doc (3 hours)
5. Identify strategic gaps (2 hours) Total: 12 hours, done sequentially

Orchestrated Process (New):

• Agent A (Research): Scrape competitors' product pages, extract features
• Agent B (Launch Tracker): Monitor competitors' release notes, blog posts, social media
• Agent C (Pricing Analyzer): Track pricing pages, identify changes
• Agent D (Synthesis): Compile findings from A, B, C into structured report
• Agent E (Gap Analysis): Compare our roadmap vs. competitor features, flag gaps Total: Agents A-C run in parallel (3 hours), then D and E run sequentially (2 hours) = 5 hours + 1 hour PM review = 6 hours total (50% time saved)

Agent offers:

I'll help you decompose your task. Describe your current process step-by-step, and I'll identify:

• Which steps can be agent-handled
• Which steps can run in parallel
• Which steps require human judgment

User response: [Describe process]

Agent provides: Decomposed workflow with agent assignments.

Step 3: Design Parallel vs. Sequential Flows

Agent asks:

Now let's design the workflow topology—which agents run in parallel vs. sequentially.

Three Flow Patterns:

Pattern 1: Full Parallel (Fastest)

[Agent A] ─┐
[Agent B] ─┼─→ [PM Review] → [Decision]
[Agent C] ─┘

• All agents run simultaneously
• PM reviews all outputs together
• Use when: Agents are independent (no dependencies between them)

Pattern 2: Pipeline (Sequential with Handoffs)

[Agent A] → [Agent B] → [Agent C] → [PM Review] → [Decision]

• Each agent depends on previous agent's output
• Use when: Later agents need earlier agents' results

Pattern 3: Hybrid (Parallel + Sequential)

[Agent A] ─┐
[Agent B] ─┼─→ [Agent D] → [Agent E] → [PM Review] → [Decision]
[Agent C] ─┘

• Some agents run in parallel, feed into sequential agents
• Use when: Early steps are independent, later steps synthesize

For your task, which pattern fits best?

Options:

1. Full Parallel — All agents are independent
2. Pipeline — Each agent needs the previous one's output
3. Hybrid — Mix of parallel and sequential
4. Not sure — [Agent analyzes your decomposition and recommends]

User response: [Selection]

Agent provides: Visual workflow diagram with agent dependencies.

Step 4: Define Agent Boundaries and Handoffs

Agent asks:

Now let's define what each agent knows (context boundaries) and how agents hand off to each other.

For each agent, define:

1. Agent Name & Purpose (What's its job?)
2. Context Needed (What information does it need?)
   • Always-available (from context engineering)
   • Retrieved on-demand (specific to this run)
3. Output Format (What does it produce?)
4. Handoff to Next Agent (What gets passed? What gets filtered?)

Example (Competitive Intelligence Orchestration):

Agent A: Competitor Research Agent

• Purpose: Scrape 10 competitors' product pages, extract features
• Context Needed:
  • Always-available: List of competitors (from constraints registry)
  • Retrieved: Competitor URLs (from database)
• Output Format: Structured JSON: {competitor, features[], last_updated}
• Handoff: Passes JSON to Agent D (Synthesis Agent)

Agent D: Synthesis Agent

• Purpose: Compile findings from Agents A, B, C into structured report
• Context Needed:
  • Always-available: Report template, writing tone (from context workspace)
  • Retrieved: Outputs from Agents A, B, C
• Output Format: Markdown document with sections: Overview, Key Features, Pricing, Launch Activity
• Handoff: Passes report to Agent E (Gap Analysis Agent)

Agent offers:

I'll help you define boundaries for each agent. For your workflow, let's define:

• What context each agent needs (always-available vs. retrieved)
• What format each agent produces (JSON, Markdown, structured data)
• How agents hand off (what's passed vs. filtered)

User response: [Describe agent needs]

Agent provides: Complete agent boundary definitions + handoff specifications.

Step 5: Implement Strategic Intent Alignment

Agent asks:

To prevent agents from "going rogue," we need to feed them the correct mix of mission, constraints, and priorities.

Strategic Intent Checklist (Per Agent):

1. Mission: What is this agent's primary goal?

   • Example: "Identify competitive threats that could impact Q2 roadmap"

2. Constraints: What must the agent NOT do?

   • Example: "Do not recommend features outside our strategic focus areas"
   • Example: "Do not scrape data from sites that prohibit it (check robots.txt)"

3. Priorities: If the agent must choose, what matters most?

   • Example: "Prioritize features launched in last 30 days over older features"

4. Evidence Standards: What counts as valid information?

   • Example: "Only cite information from official product pages, release notes, or verified blog posts"

Connection: This is context engineering applied to orchestration. See context-engineering-advisor for deeper guidance.

Agent offers:

For each agent in your workflow, let's define:

• Mission (what's the goal?)
• Constraints (what's out of bounds?)
• Priorities (what matters most?)
• Evidence standards (what counts as valid?)

User response: [Define strategic intent per agent]

Agent provides: Strategic intent document for each agent.

Step 6: Set Up Launch Control Tower Monitoring

Agent asks:

The Launch Control Tower provides real-time monitoring to flag gaps before they become critical failures.

What to Monitor (Three Dimensions):

1. Agent Performance:

• Are agents completing tasks on time?
• Are outputs meeting quality standards?
• Are agents staying within token/cost budgets?

2. Organizational Readiness (If Launching a Feature):

• Support readiness: Docs updated? Training complete? Escalation paths defined?
• Marketing readiness: Messaging finalized? Assets created? GTM plan locked?
• Operations readiness: Infrastructure scaled? Monitoring in place? Rollback plan ready?

3. Risk Flags:

• Are agents producing unexpected outputs? (potential hallucination)
• Are agents violating constraints? (ethical/compliance issues)
• Are handoffs failing? (Agent B not receiving Agent A's output correctly)

Monitoring Approach:

Option 1: Manual Dashboard

• Weekly PM review of agent outputs
• Spot-check quality, compare to golden datasets
• Manual risk flag identification

Option 2: Automated Monitoring

• Agents log outputs to central dashboard
• Automated evals run on each output (see Step 7)
• Alerts triggered when quality drops below threshold

Option 3: Hybrid

• Automated evals + weekly PM review
• Alerts for critical issues, PM spot-checks others

Which monitoring approach fits your team's maturity?

Options:

1. Manual Dashboard (Lower maturity, smaller scale)
2. Automated Monitoring (Higher maturity, larger scale)
3. Hybrid (Recommended for most teams)

User response: [Selection]

Agent provides: Monitoring plan with specific metrics and alert thresholds.

Step 7: Create Evaluation Plan

Agent asks:

To ensure orchestrated workflows produce high-quality outputs, we need an evaluation framework.

The Four Eval Types (From Research):

Evaluation Process:

1. Collect Traces: Log inputs and outputs for each agent run
2. Review Weekly: Annotate traces to categorize failure modes
3. Build Failure Mode Taxonomy: Common errors (hallucination, constraint violation, low quality)
4. Create Automated Evals: Target common failure modes with Golden Datasets or Code Assertions
5. Iterate: Refine agent prompts/context based on eval results

For your workflow, which evals make sense?

Options:

1. Start with Golden Datasets — Create 20 examples of good outputs for each agent
2. Start with Code Assertions — Define structural rules (e.g., output must be valid JSON with required fields)
3. Start with Human Evals — PM reviews every output for first 2 weeks, identifies patterns
4. All three — Comprehensive eval strategy (recommended for critical workflows)

User response: [Selection]

Agent provides: Evaluation plan with specific eval types, frequencies, and success criteria.

Step 8: Implementation Plan & Next Steps

Agent synthesizes:

Here's your complete orchestrated workflow plan:

┌─────────────────────────────────────────────────────────────┐
│ ORCHESTRATED WORKFLOW: [Your Task Name]                    │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│ TOPOLOGY: [Full Parallel / Pipeline / Hybrid]              │
│                                                             │
│ AGENTS:                                                     │
│ • Agent A: [Name] - [Purpose]                              │
│ • Agent B: [Name] - [Purpose]                              │
│ • Agent C: [Name] - [Purpose]                              │
│                                                             │
│ FLOW:                                                       │
│ [Agent A] ─┐                                                │
│ [Agent B] ─┼─→ [Agent D] → [PM Review] → [Decision]        │
│ [Agent C] ─┘                                                │
│                                                             │
│ CONTEXT BOUNDARIES:                                         │
│ • Always-available: [List]                                  │
│ • Retrieved on-demand: [List]                               │
│                                                             │
│ STRATEGIC INTENT:                                           │
│ • Mission: [Per agent]                                      │
│ • Constraints: [Per agent]                                  │
│ • Priorities: [Per agent]                                   │
│                                                             │
│ MONITORING:                                                 │
│ • Approach: [Manual / Automated / Hybrid]                   │
│ • Metrics: [List]                                           │
│ • Alert thresholds: [List]                                  │
│                                                             │
│ EVALUATION:                                                 │
│ • Golden Datasets: [Y/N]                                    │
│ • Code Assertions: [Y/N]                                    │
│ • Human Evals: [Frequency]                                  │
│                                                             │
│ TIME SAVINGS:                                               │
│ • Manual (old): [X hours]                                   │
│ • Orchestrated (new): [Y hours]                             │
│ • Savings: [X-Y hours per week]                             │
└─────────────────────────────────────────────────────────────┘

Implementation Roadmap:

Week 1: Build Context Foundations

• If not already done, complete context-engineering-advisor
• Create constraints registry, operational glossary
• Document strategic intent for each agent

Week 2: Implement First Agent

• Start with simplest agent (usually research or data collection)
• Test with 5 real examples
• Validate output quality

Week 3: Add Remaining Agents

• Implement agents 2-5 (depending on your workflow)
• Test handoffs between agents
• Validate full workflow end-to-end

Week 4: Set Up Monitoring & Evals

• Implement chosen monitoring approach
• Create Golden Datasets (if selected)
• Run first week of human evals, identify failure modes

Week 5+: Iterate & Scale

• Refine agent prompts based on eval results
• Add automated evals targeting common failures
• Scale to more use cases

Success Criteria:

• ✅ Time savings: 30%+ reduction in manual work
• ✅ Quality: Agent outputs meet quality bar 80%+ of time
• ✅ Consistency: Outputs follow template/format 95%+ of time
• ✅ Monitoring: Alerts catch issues before PM notices
• ✅ Evaluation: Failure mode taxonomy documented, automated evals in place

Agent offers:

Would you like me to:

1. Export this workflow plan as a Markdown document?
2. Provide implementation templates (agent prompt templates, monitoring dashboard, eval checklists)?
3. Recommend tools for building this orchestration (Claude Projects, custom APIs, orchestration platforms)?

Examples

Example 1: Competitive Intelligence Orchestration

Context:

• PM at B2B SaaS company
• Tracks 10 competitors
• Manual process: 12 hours per week (research, synthesis, gap analysis)

Goal: Reduce to 6 hours with orchestrated workflow

Workflow Design:

Agents:

• Agent A (Research): Scrapes competitors' product pages, extracts features
• Agent B (Launch Tracker): Monitors release notes, blogs, social media for new launches
• Agent C (Pricing Analyzer): Tracks pricing pages, identifies changes
• Agent D (Synthesis): Compiles findings from A, B, C into structured report
• Agent E (Gap Analysis): Compares our roadmap vs. competitors, flags strategic gaps

Topology: Hybrid (A, B, C in parallel → D → E → PM Review)

Strategic Intent:

• Mission (Agent E): "Identify competitive threats that could impact Q2 roadmap"
• Constraint (Agent E): "Do not recommend features outside our strategic focus areas (enterprise security, AI-powered analytics)"
• Priority (Agent B): "Prioritize features launched in last 30 days"

Monitoring:

• Hybrid approach: Automated evals + weekly PM review
• Metrics: Agent completion time, output quality score (LLM-as-Judge)
• Alert: If Agent E flags >5 critical gaps (potential strategic threat)

Evaluation:

• Golden Datasets: 20 examples of well-synthesized competitive reports
• Code Assertions: Output must be valid JSON with required fields (competitor, features[], pricing[], launch_date)
• Human Evals: PM reviews one full workflow run per week

Result:

• Time: 12 hours → 6 hours (50% savings)
• Quality: Consistent format, no missed competitors
• Strategic value: Gap analysis flags threats 2 weeks earlier than manual process

Example 2: Customer Feedback Synthesis

Context:

• PM at consumer app
• Receives 200+ pieces of feedback per week (app reviews, support tickets, NPS comments)
• Manual process: 8 hours per week (reading, tagging, synthesizing themes)

Goal: Reduce to 2 hours with orchestrated workflow

Workflow Design:

Agents:

• Agent A (Ingestion): Collects feedback from App Store, Google Play, Zendesk, NPS surveys
• Agent B (Tagging): Tags feedback by category (bug, feature request, usability, performance)
• Agent C (Sentiment): Classifies sentiment (positive, neutral, negative, critical)
• Agent D (Theme Extraction): Identifies recurring themes across feedback
• Agent E (Prioritization): Scores themes by frequency + sentiment intensity
• Agent F (Synthesis): Generates weekly synthesis report with top 5 themes + example quotes

Topology: Pipeline (A → B → C → D → E → F → PM Review)

Strategic Intent:

• Mission (Agent D): "Identify unmet customer needs that could become roadmap opportunities"
• Constraint (Agent E): "Do not deprioritize critical bugs, even if infrequent"
• Priority (Agent F): "Focus on themes affecting 10+ users in last 7 days"

Monitoring:

• Automated: Dashboard shows agent completion status, error rate
• Alerts: If Agent C flags >10 "critical" sentiment items (escalate to PM immediately)

Evaluation:

• Golden Datasets: 50 examples of well-tagged, well-synthesized feedback
• LLM-as-Judge: Weekly eval of Agent D's theme extraction quality
• Human Evals: PM spot-checks 10% of tagged feedback

Result:

• Time: 8 hours → 2 hours (75% savings)
• Quality: Themes now backed by quantitative data (frequency, sentiment)
• Strategic value: Identified 3 high-impact themes that became Q2 roadmap features

Example 3: Risk Analysis at Scale

Context:

• PM at enterprise platform
• Needs to analyze 50 features for risk (security, compliance, performance, usability)
• Manual process: 15 hours (30 minutes per feature)

Goal: Reduce to 5 hours with orchestrated workflow

Workflow Design:

Agents:

• Agent A (Security Risk): Analyzes feature for security vulnerabilities (data exposure, auth issues)
• Agent B (Compliance Risk): Checks feature against regulatory requirements (GDPR, HIPAA, SOC2)
• Agent C (Performance Risk): Estimates load impact, identifies scaling concerns
• Agent D (Usability Risk): Flags complex workflows, accessibility issues
• Agent E (Synthesis): Compiles risk scores, generates risk matrix

Topology: Full Parallel (A, B, C, D run simultaneously → E → PM Review)

Strategic Intent:

• Mission (All agents): "Flag risks that could delay launch or harm customers"
• Constraint (Agent B): "Must cite specific regulatory requirements (not vague warnings)"
• Priority (Agent E): "Prioritize risks by impact × likelihood (standard risk matrix)"

Monitoring:

• Automated: Dashboard shows risk distribution (low/medium/high/critical)
• Alerts: If any feature scores "critical" risk (escalate to PM + leadership)

Evaluation:

• Golden Datasets: 30 examples of well-analyzed features with known risks
• Code Assertions: Risk scores must be {low, medium, high, critical}—no other values
• Human Evals: Security team reviews Agent A outputs monthly

Result:

• Time: 15 hours → 5 hours (67% savings)
• Quality: Consistent risk scoring across all 50 features
• Strategic value: Identified 3 critical risks that would have been missed in manual review

Common Pitfalls

1. Orchestrating Before Context Engineering

Failure Mode: Building agent workflows without context foundations (constraints, glossary, evidence standards).

Consequence: Agents produce inconsistent outputs, violate constraints, hallucinate.

Fix: Complete context-engineering-advisor first. Build constraints registry, operational glossary, strategic intent documents.

2. Over-Orchestrating Simple Tasks

Failure Mode: Creating complex multi-agent workflows for tasks that take <2 hours per week.

Consequence: Orchestration overhead (setup, monitoring, maintenance) exceeds time saved.

Fix: Only orchestrate tasks that take 5+ hours per week or require consistency at scale.

3. No Evaluation Plan

Failure Mode: "Set it and forget it"—agents run without quality checks.

Consequence: Quality drift over time, unnoticed hallucinations, constraint violations.

Fix: Implement Golden Datasets + weekly Human Evals at minimum. Build failure mode taxonomy, create automated evals.

4. Ignoring Handoff Failures

Failure Mode: Assuming agents will correctly pass data to each other without testing.

Consequence: Agent B receives malformed data from Agent A, produces garbage output.

Fix: Test handoffs explicitly. Validate data format at each handoff. Use Code Assertions to enforce structure.

5. Forgetting the "PM Review" Step

Failure Mode: Treating orchestrated workflows as fully autonomous—no human oversight.

Consequence: Agents make decisions that lack context, empathy, or strategic alignment.

Fix: Always include PM Review step. Agents generate hypotheses/recommendations; PM validates and decides.

References

Related Skills

• context-engineering-advisor (Interactive) — Prerequisite: Build context foundations before orchestration
• ai-shaped-readiness-advisor (Interactive) — Assesses Agent Orchestration maturity (Competency #2)
• pol-probe-advisor (Interactive) — Vibe Coding connection (rapid prototyping)
• discovery-process (Workflow) — Use orchestrated agents in discovery phase

External Frameworks

• Dean Peters — The Product Manager as an Orchestrator (Productside research)
• Dean Peters — Productside Blueprint (Four AI Management Workflows: Context Engineering, Synthetic Evals, Agentic Workflows, Vibe Coding)
• Teresa Torres — Continuous Discovery Habits (5 AI PM disciplines: Context Engineering, Orchestration, Observability, Evals, Maintenance)
• Marty Cagan — Empowered (4 big risks: Value, Usability, Feasibility, Viability)

Tools & Platforms

• Claude Projects — Simple orchestration with multiple agents in one project
• LangChain — Framework for building agent chains
• LangGraph — State machine orchestration for complex workflows
• n8n / Zapier — No-code workflow automation (simpler orchestrations)

Purpose

Assess whether your product work is "AI-first" (using AI to automate existing tasks faster) or "AI-shaped" (fundamentally redesigning how product teams operate around AI capabilities). Use this to evaluate your readiness across 5 essential PM competencies for 2026, identify gaps, and get concrete recommendations on which capability to build first.

Key Distinction: AI-first is cute (using Copilot to write PRDs faster). AI-shaped is survival (building a durable "reality layer" that both humans and AI trust, orchestrating AI workflows, compressing learning cycles).

This is not about AI tools—it's about organizational redesign around AI as co-intelligence. The interactive skill guides you through a maturity assessment, then recommends your next move.

Input

Works best with: A description of how your team currently uses AI in its product work — even 'barely' is a valid answer. Also useful: Team size, product domain, and which of the 5 competencies you suspect is weakest.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The advisor opens by asking how AI currently shows up in your team's day-to-day product work.

Example invocation: Assess my team: 6 PMs, we use ChatGPT for PRD drafts and meeting summaries but nothing in our discovery or delivery process has changed.

Key Concepts

AI-First vs. AI-Shaped

Critical Insight: If a competitor can replicate your AI usage by throwing bodies at it, it's not differentiation—it's just efficiency (which becomes table stakes within months).

The 5 Essential PM Competencies (2026)

These competencies define AI-shaped product work. You'll assess your maturity on each.

1. Context Design

Building a durable "reality layer" that both humans and AI can trust—treating AI attention as a scarce resource and allocating it deliberately.

What it includes:

• Documenting what's true vs. assumed
• Immutable constraints (technical, regulatory, strategic)
• Operational glossary (shared definitions)
• Evidence standards (what counts as validation)
• Context boundaries (what to persist vs. retrieve)
• Memory architecture (short-term conversational + long-term persistent)
• Retrieval strategies (semantic search, contextual retrieval)

Key Principle: "If you can't point to evidence, constraints, and definitions, you don't have context. You have vibes."

Critical Distinction: Context Stuffing vs. Context Engineering

• Context Stuffing (AI-first): Jamming volume without intent ("paste entire PRD")
• Context Engineering (AI-shaped): Shaping structure for attention (bounded domains, retrieve with intent)

The 5 Diagnostic Questions:

1. What specific decision does this support?
2. Can retrieval replace persistence?
3. Who owns the context boundary?
4. What fails if we exclude this?
5. Are we fixing structure or avoiding it?

AI-first version: Pasting PRDs into ChatGPT; no context boundaries; "more is better" mentality AI-shaped version: CLAUDE.md files, evidence databases, constraint registries AI agents reference; two-layer memory architecture; Research→Plan→Reset→Implement cycle to prevent context rot

Deep Dive: See context-engineering-advisor for detailed guidance on diagnosing context stuffing and implementing memory architecture.

2. Agent Orchestration

Creating repeatable, traceable AI workflows (not one-off prompts).

What it includes:

• Defined workflow loops: research → synthesis → critique → decision → log rationale
• Each step shows its work (traceable reasoning)
• Workflows run consistently (same inputs = predictable process)
• Version-controlled prompts and agents

Key Principle: One-off prompts are tactical. Orchestrated workflows are strategic.

AI-first version: "Ask ChatGPT to analyze this user feedback" AI-shaped version: Automated workflow that ingests feedback, tags themes, generates hypotheses, flags contradictions, logs decisions

3. Outcome Acceleration

Using AI to compress learning cycles (not just speed up tasks).

What it includes:

• Eliminate validation lag (PoL probes run in days, not weeks)
• Remove approval delays (AI pre-validates against constraints)
• Cut meeting overhead (async AI synthesis replaces status meetings)

Key Principle: Do less, purposefully. AI removes bottlenecks, not generates more work.

AI-first version: "AI writes user stories faster" AI-shaped version: "AI runs feasibility checks overnight, eliminating 2 weeks of technical discovery"

4. Team-AI Facilitation

Redesigning team systems so AI operates as co-intelligence, not an accountability shield.

What it includes:

• Review norms (who checks AI outputs, when, how)
• Evidence standards (AI must cite sources, not hallucinate)
• Decision authority (AI recommends, humans decide—clear boundaries)
• Psychological safety (team can challenge AI without feeling "dumb")

Key Principle: AI amplifies judgment, doesn't replace accountability.

AI-first version: "I used AI" as excuse for bad outputs AI-shaped version: Clear review protocols; AI outputs treated as drafts requiring human validation

5. Strategic Differentiation

Moving beyond efficiency to create defensible competitive advantages.

What it includes:

• New customer capabilities (what can users do now that they couldn't before?)
• Workflow rewiring (processes competitors can't replicate without full redesign)
• Economics competitors can't match (10x cost advantage through AI)

Key Principle: "If a competitor can copy it by throwing bodies at it, it's not differentiation."

AI-first version: "We use AI to write better docs" AI-shaped version: "We validate product hypotheses in 2 days vs. industry standard 3 weeks—ship 6x more validated features per quarter"

Anti-Patterns (What This Is NOT)

• Not about AI tools: Using Claude vs. ChatGPT doesn't matter. Redesigning workflows matters.
• Not about speed: Writing PRDs 2x faster isn't strategic if PRDs weren't the bottleneck.
• Not about automation: Automating bad processes just scales the bad.
• Not about replacing humans: AI-shaped orgs augment judgment, not eliminate it.

When to Use This Skill

✅ Use this when:

• You're using AI tools but not seeing strategic advantage
• You suspect you're "AI-first" (efficiency) but want to be "AI-shaped" (transformation)
• You need to prioritize which AI capability to build next
• Leadership asks "How are we using AI?" and you're not sure how to answer strategically
• You want to assess team readiness for AI-powered product work

❌ Don't use this when:

• You haven't started using AI at all (start with basic tools first)
• You're looking for tool recommendations (this is about organizational design, not tooling)
• You need tactical "how to write a prompt" guidance (use skills for that)

Facilitation Source of Truth

Use workshop-facilitation as the default interaction protocol for this skill.

It defines:

• session heads-up + entry mode (Guided, Context dump, Best guess)
• one-question turns with plain-language prompts
• progress labels (for example, Context Qx/8 and Scoring Qx/5)
• interruption handling and pause/resume behavior
• numbered recommendations at decision points
• quick-select numbered response options for regular questions (include Other (specify) when useful)

This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.

Application

This interactive skill uses adaptive questioning to assess your maturity across 5 competencies, then recommends which to prioritize.

Facilitation Protocol (Mandatory)

1. Ask exactly one question per turn.
2. Wait for the user's answer before asking the next question.
3. Use plain-language questions (no shorthand labels as the primary question). If needed, include an example response format.
4. Show progress on every turn using user-facing labels:
   • Context Qx/8 during context gathering
   • Scoring Qx/5 during maturity scoring
   • Include "questions remaining" when practical.
5. Do not use internal phase labels (like "Step 0") in user-facing prompts unless the user asks for internal structure details.
6. For maturity scoring questions, present concise 1-4 choices first; share full rubric details only if requested.
7. For context questions, offer concise numbered quick-select options when practical, plus Other (specify) for open-ended answers. Accept multi-select replies like 1,3 or 1 and 3.
8. Give numbered recommendations only at decision points, not after every answer.
9. Decision points include:
   • After the full context summary
   • After the 5-dimension maturity profile
   • During priority selection and action-plan path selection
10. When recommendations are shown, enumerate clearly (1., 2., 3.) and accept selections like #1, 1, 1 and 3, 1,3, or custom text.
11. If multiple options are selected, synthesize a combined path and continue.
12. If custom text is provided, map it to the closest valid path and continue without forcing re-entry.
13. Interruption handling is mandatory: if the user asks a meta question ("how many left?", "why this label?", "pause"), answer directly first, then restate current progress and resume with the pending question.
14. If the user says to stop or pause, halt the assessment immediately and wait for explicit resume.
15. If the user asks for "one question at a time," keep that mode for the rest of the session unless they explicitly opt out.
16. Before any assessment question, give a short heads-up on time/length and let the user choose an entry mode.

Session Start: Heads-Up + Entry Mode (Mandatory)

Agent opening prompt (use this first):

"Quick heads-up before we start: this usually takes about 7-10 minutes and up to 13 questions total (8 context + 5 scoring).

How do you want to do this?

1. Guided mode: I’ll ask one question at a time.
2. Context dump: you paste what you already know, and I’ll skip anything redundant.
3. Best guess mode: I’ll make reasonable assumptions where details are missing, label them, and keep moving."

Accept selections as #1, 1, 1 and 3, 1,3, or custom text.

Mode behavior:

• If Guided mode: Run Step 0 as written, then scoring.
• If Context dump: Ask for pasted context once, summarize it, identify gaps, and:
  • Skip any context questions already answered.
  • Ask only the minimum missing context needed (0-2 clarifying questions).
  • Move to scoring as soon as context is sufficient.
• If Best guess mode: Ask for the smallest viable starting input (role/team + primary goal), then:
  • Infer missing details using reasonable defaults.
  • Label each inferred item as Assumption.
  • Include confidence tags (High, Medium, Low) for each assumption.
  • Continue without blocking on unknowns.

At the final summary, include an Assumptions to Validate section when context dump or best guess mode was used.

Step 0: Gather Context

Agent asks:

Collect context using this exact sequence, one question at a time:

1. "Which AI tools are you using today?"
2. "How does your team usually use AI today: one-off prompts, reusable templates, or multi-step workflows?"
3. "Who uses AI consistently today: just you, PMs, or cross-functional teams?"
4. "About how many PMs, engineers, and designers are on your team?"
5. "What stage are you in: startup, growth, or enterprise?"
6. "How are decisions made: centralized, distributed, or consensus-driven?"
7. "What competitive advantage are you trying to build with AI?"
8. "What's the biggest bottleneck slowing learning and iteration today?"

After question 8, summarize back in 4 lines:

• Current AI usage pattern
• Team context
• Strategic intent
• Primary bottleneck

Step 1: Context Design Maturity

Agent asks:

Let's assess your Context Design capability—how well you've built a "reality layer" that both humans and AI can trust, and whether you're doing context stuffing (volume without intent) or context engineering (structure for attention).

Which statement best describes your current state?

1. Level 1 (AI-First / Context Stuffing): "I paste entire documents into ChatGPT every time I need something. No shared knowledge base. No context boundaries."

   • Reality: One-off prompting with no durability; "more is better" mentality
   • Problem: AI has no memory; you repeat yourself constantly; context stuffing degrades attention
   • Context Engineering Gap: No answers to the 5 diagnostic questions; persisting everything "just in case"

2. Level 2 (Emerging / Early Structure): "We have some docs (PRDs, strategy memos), but they're scattered. No consistent format. Starting to notice context stuffing issues (vague responses, normalized retries)."

   • Reality: Context exists but isn't structured for AI consumption; no retrieval strategy
   • Problem: AI can't reliably find or trust information; mixing always-needed with episodic context
   • Context Engineering Gap: No context boundary owner; no distinction between persist vs. retrieve

3. Level 3 (Transitioning / Context Engineering Emerging): "We've started using CLAUDE.md files and project instructions. Constraints registry exists. We're identifying what to persist vs. retrieve. Experimenting with Research→Plan→Reset→Implement cycle."

   • Reality: Structured context emerging, but not comprehensive; context boundaries defined but not fully enforced
   • Problem: Coverage is patchy; some areas well-documented, others vibe-driven; inconsistent retrieval practices
   • Context Engineering Progress: Can answer 3-4 of the 5 diagnostic questions; context boundary owner assigned; starting to use two-layer memory

4. Level 4 (AI-Shaped / Context Engineering Mastery): "We maintain a durable reality layer: constraints registry (20+ entries), evidence database, operational glossary (30+ terms). Two-layer memory architecture (short-term conversational + long-term persistent via vector DB). Context boundaries defined and owned. AI agents reference these automatically. We use Research→Plan→Reset→Implement to prevent context rot."

   • Reality: Comprehensive, version-controlled context both humans and AI trust; retrieval with intent (not completeness)
   • Outcome: AI operates with high confidence; reduces hallucination and rework; token usage optimized; no context stuffing
   • Context Engineering Mastery: Can answer all 5 diagnostic questions; context boundary audited quarterly; quantitative efficiency tracking: (Accuracy × Coherence) / (Tokens × Latency)

Select your level: [1, 2, 3, or 4]

Note: If you selected Level 1-2 and struggle with context stuffing, consider using context-engineering-advisor to diagnose and fix Context Hoarding Disorder before proceeding.

User response: [Selection]

Agent records: Context Design maturity = [Level X]

Step 2: Agent Orchestration Maturity

Agent asks:

Now let's assess Agent Orchestration—whether you have repeatable AI workflows or just one-off prompts.

Which statement best describes your current state?

1. Level 1 (AI-First): "I type prompts into ChatGPT as needed. No saved workflows or templates."

   • Reality: Tactical, ad-hoc usage
   • Problem: Inconsistent results; can't scale or audit

2. Level 2 (Emerging): "I have a few saved prompts I reuse. Maybe some custom GPTs or Claude Projects."

   • Reality: Repeatable prompts, but not full workflows
   • Problem: Each step is manual; no orchestration

3. Level 3 (Transitioning): "We've built some multi-step workflows (research → synthesis → critique). Tracked in tools like Notion or Linear."

   • Reality: Workflows exist but require manual handoffs
   • Problem: Still human-in-the-loop for every step; not fully automated

4. Level 4 (AI-Shaped): "We have orchestrated AI workflows that run autonomously: research → synthesis → critique → decision → log rationale. Each step is traceable and version-controlled."

   • Reality: Workflows run consistently; show their work at each step
   • Outcome: Reliable, auditable, scalable AI processes

Select your level: [1, 2, 3, or 4]

User response: [Selection]

Agent records: Agent Orchestration maturity = [Level X]

Step 3: Outcome Acceleration Maturity

Agent asks:

Next, Outcome Acceleration—are you using AI to compress learning cycles, or just speed up tasks?

Which statement best describes your current state?

1. Level 1 (AI-First): "AI helps me write docs faster (PRDs, user stories). Saves me a few hours per week."

   • Reality: Efficiency gains on artifact creation
   • Problem: Docs weren't the bottleneck; learning cycles unchanged

2. Level 2 (Emerging): "AI helps with research and synthesis (summarize user feedback, analyze competitors). Saves research time."

   • Reality: Modest learning acceleration
   • Problem: Still sequential; AI doesn't eliminate validation lag

3. Level 3 (Transitioning): "We use AI to run experiments faster (PoL probes, feasibility checks). Cut validation time from weeks to days."

   • Reality: Learning cycles compressing
   • Problem: Not yet systematic; only applied to some experiments

4. Level 4 (AI-Shaped): "AI systematically removes bottlenecks: overnight feasibility checks, async synthesis replaces meetings, automated validation against constraints. Learning cycles 5-10x faster."

   • Reality: Fundamental redesign of how learning happens
   • Outcome: Ship validated features 6x faster than competitors

Select your level: [1, 2, 3, or 4]

User response: [Selection]

Agent records: Outcome Acceleration maturity = [Level X]

Step 4: Team-AI Facilitation Maturity

Agent asks:

Now assess Team-AI Facilitation—how well you've redesigned team systems for AI as co-intelligence.

Which statement best describes your current state?

1. Level 1 (AI-First): "I use AI privately. Team doesn't know or doesn't use it. No shared norms."

   • Reality: Individual tool usage, no team integration
   • Problem: Inconsistent quality; no accountability for AI outputs

2. Level 2 (Emerging): "Team uses AI, but no formal review process. 'I used AI' mentioned casually."

   • Reality: Awareness but no structure
   • Problem: AI outputs treated as final; errors slip through

3. Level 3 (Transitioning): "We have review norms emerging (AI outputs are drafts, not finals). Evidence standards discussed but not codified."

   • Reality: Cultural shift underway
   • Problem: Norms are informal; not everyone follows them

4. Level 4 (AI-Shaped): "Clear protocols: AI outputs require human validation, evidence standards codified, decision authority explicit (AI recommends, humans decide). Team treats AI as co-intelligence."

   • Reality: AI integrated into team operating system
   • Outcome: High-quality outputs; psychological safety maintained

Select your level: [1, 2, 3, or 4]

User response: [Selection]

Agent records: Team-AI Facilitation maturity = [Level X]

Step 5: Strategic Differentiation Maturity

Agent asks:

Finally, Strategic Differentiation—are you creating defensible competitive advantages, or just efficiency gains?

Which statement best describes your current state?

1. Level 1 (AI-First): "We use AI to work faster (write better docs, respond to customers quicker). Efficiency gains only."

   • Reality: Table-stakes improvements
   • Problem: Competitors can copy this within months

2. Level 2 (Emerging): "AI enables us to do things we couldn't before (analyze 10x more data, test more hypotheses). New capabilities, but competitors could replicate."

   • Reality: Capability expansion, but not defensible
   • Problem: No moat; competitors hire more people to match

3. Level 3 (Transitioning): "We've redesigned some workflows around AI (e.g., validate hypotheses in 2 days vs. 3 weeks). Starting to create separation."

   • Reality: Workflow advantages emerging
   • Problem: Not yet systematic; only applied in pockets

4. Level 4 (AI-Shaped): "We've fundamentally rewired how we operate: customers get capabilities they can't get elsewhere, our learning cycles are 10x faster than industry standard, our economics are 5x better. Competitors can't replicate without full org redesign."

   • Reality: Defensible competitive moat
   • Outcome: Strategic advantage that compounds over time

Select your level: [1, 2, 3, or 4]

User response: [Selection]

Agent records: Strategic Differentiation maturity = [Level X]

Step 6: Assess Maturity Profile

Agent synthesizes:

Here's your AI-Shaped Readiness Profile:

┌─────────────────────────────┬───────┬──────────┐
│ Competency                  │ Level │ Maturity │
├─────────────────────────────┼───────┼──────────┤
│ 1. Context Design           │   X   │ [Label]  │
│ 2. Agent Orchestration      │   X   │ [Label]  │
│ 3. Outcome Acceleration     │   X   │ [Label]  │
│ 4. Team-AI Facilitation     │   X   │ [Label]  │
│ 5. Strategic Differentiation│   X   │ [Label]  │
└─────────────────────────────┴───────┴──────────┘

Overall Assessment: [AI-First / Emerging / Transitioning / AI-Shaped]

Maturity Labels:

• Level 1: AI-First (efficiency only)
• Level 2: Emerging (early capabilities)
• Level 3: Transitioning (redesign underway)
• Level 4: AI-Shaped (strategic transformation)

Overall Assessment Logic:

• AI-First: Average level 1-1.5 (mostly Level 1s)
• Emerging: Average level 2-2.5 (mostly Level 2s)
• Transitioning: Average level 3-3.5 (mostly Level 3s)
• AI-Shaped: Average level 3.5-4 (mostly Level 4s)

Step 7: Identify Priority Gap

Agent asks:

Based on your maturity profile, which competency should you prioritize first?

Agent analyzes dependencies:

Dependency Logic:

1. Context Design is foundational — If Level 1-2, this must be priority #1 (Agent Orchestration and Outcome Acceleration depend on it)
2. Agent Orchestration enables Outcome Acceleration — If Context Design is Level 3+, but Agent Orchestration is Level 1-2, prioritize orchestration
3. Team-AI Facilitation is parallel — Can be developed alongside others, but required for scale
4. Strategic Differentiation requires Levels 3+ on others — Don't focus here until foundational competencies are built

Agent recommends:

Based on your profile, I recommend focusing on [Competency Name] first because:

Option 1: Context Design (if Level 1-2)

• Why: Without durable context, AI operates on vibes. Every workflow will be fragile.
• Impact: Unlocks Agent Orchestration and Outcome Acceleration
• Next Steps: Build CLAUDE.md files, start constraints registry, create operational glossary

Option 2: Agent Orchestration (if Context is 3+, but Orchestration is 1-2)

• Why: You have context, but no repeatable workflows. Scaling requires orchestration.
• Impact: Turn one-off prompts into reliable, traceable workflows
• Next Steps: Document your most frequent AI workflow, version-control prompts, add traceability

Option 3: Outcome Acceleration (if Context + Orchestration are 3+)

• Why: You have infrastructure; now compress learning cycles
• Impact: Strategic advantage emerges from speed-to-learning
• Next Steps: Identify biggest bottleneck in learning cycle, design AI workflow to eliminate it

Option 4: Team-AI Facilitation (if usage is individual, not team-wide)

• Why: Can't scale if only you're AI-shaped; team must adopt
• Impact: Organizational transformation, not just individual productivity
• Next Steps: Establish review norms, codify evidence standards, create decision authority framework

Option 5: Strategic Differentiation (if all others are 3+)

• Why: You have the foundation; now build the moat
• Impact: Create defensible competitive advantage
• Next Steps: Identify workflow competitors can't replicate, design AI-enabled customer capabilities

Which would you like to focus on?

Options:

1. Accept recommendation — [Agent provides detailed action plan]
2. Choose different priority — [Agent warns about dependencies but allows override]
3. Focus on multiple simultaneously — [Agent suggests parallel tracks if feasible]

User response: [Selection]

Step 8: Generate Action Plan

Agent provides tailored action plan based on selected priority:

If Priority = Context Design

Goal: Build a durable "reality layer" that both humans and AI trust—move from context stuffing to context engineering.

Pre-Phase: Diagnose Context Stuffing (If Needed) If you're at Level 1-2, first diagnose context stuffing symptoms:

1. Run through the 5 diagnostic questions (see context-engineering-advisor)
2. Identify what you're persisting that should be retrieved
3. Assign context boundary owner
4. Create Context Manifest (what's always-needed vs. episodic)

Phase 1: Document Constraints (Week 1)

1. Create a constraints registry:
   • Technical constraints (APIs, data models, performance limits)
   • Regulatory constraints (GDPR, HIPAA, etc.)
   • Strategic constraints (we will/won't build X)
2. Apply diagnostic question #4 to each constraint: "What fails if we exclude this?"
3. Format: Structured file AI agents can parse (YAML, JSON, or Markdown with frontmatter)
4. Version control in Git

Phase 2: Build Operational Glossary (Week 2)

1. List top 20-30 terms your team uses (e.g., "user," "customer," "activation," "churn")
2. Define each unambiguously (avoid "it depends")
3. Include edge cases and exceptions
4. Add to CLAUDE.md or project instructions
5. This becomes your long-term persistent memory (Declarative Memory)

Phase 3: Establish Evidence Standards + Context Boundaries (Week 3)

1. Define what counts as validation:
   • User feedback: "X users said Y" (with quotes)
   • Analytics: "Metric Z changed by N%" (with dashboard link)
   • Competitive intel: "Competitor A launched B" (with source)
2. Reject: "I think," "We feel," "It seems like"
3. Define context boundaries using the 5 diagnostic questions:
   • What specific decision does each piece of context support?
   • Can retrieval replace persistence?
   • Who owns the context boundary?
4. Create Context Manifest document
5. Codify in team docs

Phase 4: Implement Memory Architecture + Workflows (Week 4)

1. Set up two-layer memory:
   • Short-term (conversational): Summarize/truncate older parts of conversation
   • Long-term (persistent): Constraints registry + operational glossary (consider vector database for retrieval)
2. Implement Research→Plan→Reset→Implement cycle:
   • Research: Allow chaotic context gathering
   • Plan: Synthesize into high-density SPEC.md or PLAN.md
   • Reset: Clear context window
   • Implement: Use only the plan as context
3. Update AI prompts to reference constraints registry and glossary
4. Test: Ask AI to cite constraints when making recommendations
5. Measure: % of AI outputs that cite evidence vs. hallucinate; token usage efficiency

Success Criteria:

• ✅ Constraints registry has 20+ entries
• ✅ Operational glossary has 20-30 terms
• ✅ Evidence standards documented and shared
• ✅ Context Manifest created (always-needed vs. episodic)
• ✅ Context boundary owner assigned
• ✅ Two-layer memory architecture implemented
• ✅ Research→Plan→Reset→Implement cycle tested on 1 workflow
• ✅ AI agents reference these automatically
• ✅ Token usage down 30%+ (less context stuffing)
• ✅ Output consistency up (fewer retries)

Related Skills:

• context-engineering-advisor (Interactive) — Deep dive on diagnosing context stuffing and implementing memory architecture
• problem-statement.md — Define constraints before framing problems
• epic-hypothesis.md — Evidence-based hypothesis writing

If Priority = Agent Orchestration

Goal: Turn one-off prompts into repeatable, traceable AI workflows.

Phase 1: Map Current Workflows (Week 1)

1. Pick your most frequent AI use case (e.g., "analyze user feedback")
2. Document every step you currently take:
   • Copy/paste feedback into ChatGPT
   • Ask for themes
   • Manually categorize
   • Write summary
3. Identify pain points (manual handoffs, inconsistent results)

Phase 2: Design Orchestrated Workflow (Week 2)

1. Define workflow loop:
   • Research: AI reads all feedback (structured input)
   • Synthesis: AI identifies themes (with evidence)
   • Critique: AI flags contradictions or weak signals
   • Decision: Human reviews and decides next steps
   • Log: AI records rationale and sources
2. Each step must be traceable (show sources, reasoning)

Phase 3: Build and Test (Week 3)

1. Implement workflow using:
   • Claude Projects (if simple)
   • Custom GPTs (if moderate)
   • API orchestration (if complex)
2. Run on 3 past examples; compare to manual process
3. Measure: Time saved, consistency improved, traceability added

Phase 4: Document and Scale (Week 4)

1. Version-control prompts (Git)
2. Document workflow steps for team
3. Train 2 teammates; observe results
4. Iterate based on feedback

Success Criteria:

• ✅ At least 1 workflow runs consistently (same inputs → predictable process)
• ✅ Each step is traceable (AI cites sources)
• ✅ Team can replicate workflow without your involvement

Related Skills:

• pol-probe-advisor.md — Use orchestrated workflows for validation experiments

If Priority = Outcome Acceleration

Goal: Use AI to compress learning cycles, not just speed up tasks.

Phase 1: Identify Bottleneck (Week 1)

1. Map your current learning cycle (e.g., hypothesis → experiment → analysis → decision)
2. Time each step
3. Identify slowest step (usually: validation lag, approval delays, or meeting overhead)

Phase 2: Design AI Intervention (Week 2)

1. Ask: "What if this step happened overnight?"
   • Feasibility checks: AI spike in 2 hours vs. 2 days
   • User research synthesis: AI analysis in 1 hour vs. 1 week
   • Approval pre-checks: AI validates against constraints before meeting
2. Design minimal AI workflow to eliminate bottleneck

Phase 3: Run Pilot (Week 3)

1. Test AI intervention on 1 real initiative
2. Measure cycle time: before vs. after
3. Validate quality: Did AI maintain rigor, or cut corners?

Phase 4: Scale (Week 4)

1. If successful (cycle time down 50%+, quality maintained), apply to 3 more initiatives
2. Document workflow
3. Train team

Success Criteria:

• ✅ Learning cycle compressed by 50%+ on at least 1 initiative
• ✅ Quality maintained (no shortcuts that compromise rigor)
• ✅ Team adopts the accelerated workflow

Related Skills:

• pol-probe.md — Use AI to run PoL probes faster
• discovery-process.md — Compress discovery cycles with AI

If Priority = Team-AI Facilitation

Goal: Redesign team systems so AI operates as co-intelligence, not accountability shield.

Phase 1: Establish Review Norms (Week 1)

1. Codify rule: "AI outputs are drafts, not finals"
2. Define review protocol:
   • Who reviews AI outputs? (peer, lead PM, cross-functional partner)
   • When? (before sharing externally, before decisions)
   • What to check? (accuracy, completeness, evidence citation)
3. Share with team, get buy-in

Phase 2: Set Evidence Standards (Week 2)

1. AI must cite sources (no hallucinations)
2. Reject outputs that say "I think" or "it seems"
3. Require: "According to [source], [fact]"
4. Add to team operating docs

Phase 3: Define Decision Authority (Week 3)

1. Clarify: AI recommends, humans decide
2. Document who has authority to override AI recommendations (PM, team lead, cross-functional consensus)
3. Create escalation path (what if AI and human disagree?)

Phase 4: Build Psychological Safety (Week 4)

1. Team exercise: Share an AI mistake you caught (normalize catching errors)
2. Reward critical thinking ("Good catch on that AI hallucination!")
3. Avoid: "Why didn't you just use AI?" (shaming)

Success Criteria:

• ✅ Review norms documented and followed by team
• ✅ Evidence standards codified
• ✅ Decision authority clear
• ✅ Team comfortable challenging AI outputs

Related Skills:

• problem-statement.md — Evidence-based problem framing
• epic-hypothesis.md — Testable, evidence-backed hypotheses

If Priority = Strategic Differentiation

Goal: Create defensible competitive advantages, not just efficiency gains.

Phase 1: Identify Moat Opportunities (Week 1)

1. Ask: "What could we do with AI that competitors can't replicate by adding headcount?"
   • New customer capabilities (e.g., "AI advisor suggests personalized roadmap")
   • Workflow rewiring (e.g., "Validate product ideas in 2 days vs. 3 weeks")
   • Economics shift (e.g., "Deliver enterprise features at SMB prices via AI automation")
2. List 5 candidates
3. Prioritize by defensibility (how hard to copy?)

Phase 2: Design AI-Enabled Capability (Week 2)

1. Pick top candidate
2. Design end-to-end workflow:
   • What does customer experience?
   • What does AI do behind the scenes?
   • What human judgment is required?
3. Sketch MVP (minimum viable moat)

Phase 3: Build and Test (Weeks 3-4)

1. Build prototype (can be PoL probe, not production)
2. Test with 5 customers
3. Measure: Does this create value competitors can't match?

Phase 4: Validate Moat (Week 5)

1. Ask: "How would a competitor replicate this?"
   • If answer is "hire more people," it's not a moat
   • If answer is "redesign their entire org," you have a moat
2. Document competitive analysis
3. Decide: Build full version, pivot, or kill

Success Criteria:

• ✅ Identified at least 1 AI-enabled capability competitors can't easily copy
• ✅ Validated with customers (they see the value)
• ✅ Confirmed defensibility (competitor analysis)

Related Skills:

• positioning-statement.md — Articulate your AI-driven differentiation
• jobs-to-be-done.md — Understand what customers hire your AI capabilities to do

Step 9: Track Progress (Optional)

Agent offers:

Would you like me to create a progress tracker for your AI-shaped transformation?

Tracker includes:

• Current maturity levels (baseline)
• Target maturity levels (goal state)
• Action plan milestones (from Step 8)
• Review cadence (weekly, monthly)

Options:

1. Yes, create tracker — [Agent generates Markdown checklist]
2. No, I'll track separately — [Agent provides summary]

Examples

Example 1: Early-Stage Startup (AI-First → Emerging)

Context:

• Team: 2 PMs, 5 engineers
• AI Usage: ChatGPT for writing PRDs, occasional Copilot usage
• Goal: Move faster than larger competitors

Assessment Results:

• Context Design: Level 1 (no structured context)
• Agent Orchestration: Level 1 (one-off prompts)
• Outcome Acceleration: Level 1 (docs faster, but learning cycles unchanged)
• Team-AI Facilitation: Level 2 (team uses AI, but no norms)
• Strategic Differentiation: Level 1 (efficiency only)

Recommendation: Focus on Context Design first.

Action Plan (Week 1-4):

• Week 1: Create constraints registry (10 technical constraints)
• Week 2: Build operational glossary (15 terms)
• Week 3: Establish evidence standards
• Week 4: Add context to CLAUDE.md files

Outcome: After 4 weeks, Context Design → Level 3. Unlocks Agent Orchestration next quarter.

Example 2: Growth-Stage Company (Transitioning → AI-Shaped)

Context:

• Team: 10 PMs, 50 engineers, 5 designers
• AI Usage: Claude Projects for research, custom workflows emerging
• Goal: Build defensible AI advantage before IPO

Assessment Results:

• Context Design: Level 3 (structured context, not comprehensive)
• Agent Orchestration: Level 3 (some workflows, manual handoffs)
• Outcome Acceleration: Level 2 (modest gains, not systematic)
• Team-AI Facilitation: Level 3 (norms emerging, not codified)
• Strategic Differentiation: Level 2 (new capabilities, but copyable)

Recommendation: Focus on Outcome Acceleration (foundation is solid; now compress learning cycles).

Action Plan (Week 1-4):

• Week 1: Identify bottleneck (discovery cycles take 3 weeks)
• Week 2: Design AI workflow to run overnight feasibility checks
• Week 3: Pilot on 1 initiative (cut cycle to 5 days)
• Week 4: Scale to 3 initiatives

Outcome: Learning cycles 5x faster → strategic separation from competitors → Level 4 Outcome Acceleration + Level 3 Strategic Differentiation.

Example 3: Enterprise Company (AI-First, Scattered Usage)

Context:

• Team: 50 PMs, 300 engineers
• AI Usage: Individual PMs use various tools, no consistency
• Goal: Standardize AI usage, create cross-functional workflows

Assessment Results:

• Context Design: Level 2 (docs exist, not structured for AI)
• Agent Orchestration: Level 1 (no shared workflows)
• Outcome Acceleration: Level 1 (efficiency only)
• Team-AI Facilitation: Level 1 (private usage, no norms)
• Strategic Differentiation: Level 1 (no advantage)

Recommendation: Focus on Team-AI Facilitation first (distributed team needs shared norms before building infrastructure).

Action Plan (Week 1-4):

• Week 1: Establish review norms (AI outputs are drafts)
• Week 2: Set evidence standards (AI must cite sources)
• Week 3: Define decision authority (AI recommends, leads decide)
• Week 4: Pilot with 3 teams, gather feedback

Outcome: Team-AI Facilitation → Level 3. Creates foundation for Context Design and Agent Orchestration next.

Common Pitfalls

1. Mistaking Efficiency for Differentiation

Failure Mode: "We use AI to write PRDs 2x faster—we're AI-shaped!"

Consequence: Competitors copy within 3 months; no lasting advantage.

Fix: Ask: "If a competitor threw 2x more people at this, could they match us?" If yes, it's efficiency (table stakes), not differentiation.

2. Skipping Context Design

Failure Mode: Building Agent Orchestration workflows without durable context.

Consequence: AI workflows are fragile (context changes break everything).

Fix: Context Design is foundational. Don't skip it. Build constraints registry, glossary, evidence standards first.

3. Individual Usage, Not Team Transformation

Failure Mode: "I'm AI-shaped, but my team isn't."

Consequence: Can't scale; workflows die when you're on vacation.

Fix: Prioritize Team-AI Facilitation. Shared norms > individual productivity.

4. Focusing on Tools, Not Workflows

Failure Mode: "Should we use Claude or ChatGPT?"

Consequence: Tool debates distract from organizational redesign.

Fix: Tools don't matter. Workflows matter. Focus on redesigning how work gets done, not which AI you use.

5. Speed Over Learning

Failure Mode: "AI helps us ship faster!"

Consequence: Ship the wrong thing faster (if you're not compressing learning cycles).

Fix: Outcome Acceleration is about learning faster, not building faster. Validate hypotheses in days, not weeks.

References

Related Skills

• context-engineering-advisor (Interactive) — Deep dive on Context Design competency: Diagnose context stuffing, implement memory architecture, use Research→Plan→Reset→Implement cycle
• problem-statement (Component) — Evidence-based problem framing (Context Design)
• epic-hypothesis (Component) — Testable hypotheses with evidence standards
• pol-probe-advisor (Interactive) — Use AI to compress validation cycles (Outcome Acceleration)
• discovery-process (Workflow) — Apply AI-shaped principles to discovery
• positioning-statement (Component) — Articulate your AI-driven differentiation (Strategic Differentiation)

External Frameworks

• Dean Peters — AI-First Is Cute. AI-Shaped Is Survival. (Dean Peters' Substack, 2026)
• Dean Peters — Context Stuffing Is Not Context Engineering (Dean Peters' Substack, 2026) — Deep dive on Competency #1 (Context Design)

Further Reading

• Ethan Mollick — Co-Intelligence (on AI as co-intelligence, not replacement)
• Shreyas Doshi — Twitter threads on PM judgment augmentation with AI
• Lenny Rachitsky — Newsletter interviews with AI-forward PMs

Purpose

Defines the two-axis mental model that distinguishes Director-level thinking from PM thinking: Altitude (how wide you zoom out) and Horizon (how far ahead you look). Use this to understand what actually changes in the transition, diagnose which transition zone is creating friction, and apply the Cascading Context Map when organizational direction is vague or absent.

This is not a seniority hierarchy. A PM operating at the right altitude for their role is doing excellent work. A Director operating at PM altitude is leaving their actual job undone.

Input

Nothing required — this is a reference framework you can read start to finish. Also useful: Your current situation (role, the transition friction you're feeling, a decision you're weighing). With that, the framework can be applied directly to diagnose which transition zone you're in.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. Read it as a mental model; pair it with director-readiness-advisor for guided coaching.

Example invocation: Apply the altitude-horizon framework to my situation: senior PM, just inherited two more teams, drowning in delivery detail.

Key Concepts

The Two Axes

Altitude — Scope

• PM altitude: Close to the ground. Customer problems, individual features, sprint priorities, specific team dynamics.
• Director altitude: High-level view. Product portfolio, cross-functional systems, organizational dynamics, budget allocation, market positioning.
• The shift is not about losing empathy for customers — it's about zooming out to see the entire restaurant, not just one table.

Horizon — Time

• PM horizon: Days, weeks, sprints. A quarter at most.
• Director horizon: Quarter as the starting point. Annual planning cycles, multi-year strategy, market shifts.
• Directors plan for where the product ecosystem needs to be in a year, then work backward.

The Waiter vs. Restaurant Operator

The sharpest analogy for the role shift:

The waiter excels at translating the experience of individual diners. The operator isn't ignoring diners — they're asking different questions: "Are we overspending on ingredients? Is a 75-page menu confusing customers? Do we need another server for the dinner rush?" Neither question is more important in absolute terms. They're appropriate to different roles.

Four Transition Zones

The PM → Director shift requires movement across four zones. Most people struggle with one or two more than the others — diagnosing which one is the leverage point.

Zone 1 — Thinking Altitude

• Stop: Solving individual customer problems directly
• Start: Designing systems and teams that solve classes of problems

Zone 2 — Persona Shift

• Stop: Obsessing over individual user personas and daily customer touchpoints
• Start: Thinking in buyer personas, market cohorts, organizational stakeholders, and executive dynamics

Zone 3 — Hero Syndrome Recovery

• Stop: Being the person who saves the day and earns the pat on the back
• Start: Getting satisfaction from team success — your product is your people, not the roadmap

Zone 4 — Direction Creation

• Stop: Waiting for clear direction from above before moving
• Start: Creating context cascades that translate company strategy into team clarity, even when inputs are incomplete

Named Failure Modes

Hero Syndrome What it looks like: Jumping in to solve problems directly. Staying close to the tactical work. Wanting visibility on individual wins. Why it happens: PMs are trained to be helpful and responsive. Directors get fewer pats on the back, so they regress to the old reward loop. The cost: You under-perform as a Director while over-functioning as a senior IC. Your team doesn't develop because you're in their way.

Allergic to Process What it looks like: Resisting shared structures. Letting high-performing PMs run their own playbooks independently. Why it happens: PMs naturally resist bureaucracy. Early director permissiveness can feel like "great leadership" and "trusting the team." The cost: Stakeholders across marketing, finance, and leadership can't synthesize inconsistent outputs. Without shared processes, teams become "monkeys in the room breaking glass."

People-Pleaser Leadership What it looks like: Wanting the team to like you. Avoiding hard feedback. Saying yes to stakeholder requests to preserve relationships. Why it happens: The skills that made you a great PM — listening, empathy, responsiveness — become liabilities at organizational scale. The cost: You confuse "popular" with "effective." Respect is built through clarity and hard calls, not niceness.

Instant Gratification Trap What it looks like: Reading leadership books, collecting certifications, asking "what do I need to do to get promoted?" Why it happens: PMs are good at optimization. They try to shortcut the experience requirement. The cost: Director readiness requires war stories and lived humility. You can study your way to fluency in the vocabulary, but not to readiness for the role.

Black-and-White Thinking What it looks like: "This seems like an obvious decision." "Why can't we fund both?" "Why is everything so political here?" Why it happens: PMs operate in cleaner problem spaces with clearer cause-and-effect. Director decisions involve competing constraints, limited information, and organizational dynamics. The cost: Fast decisions with low confidence create downstream chaos. The grayscale is not a failure of leadership — it's the actual terrain.

The Cascading Context Map

When organizational direction is vague or absent, Directors don't wait — they cascade.

The six steps:

1. Listen to the top-level strategy — QBRs, company messaging, executive communications
2. Extract key priorities leadership stated — Identify 3–5 themes, not 20 bullet points
3. Map the second layer: "How does our business unit accomplish these objectives?"
4. Map the third layer: "How does our product portfolio accomplish that?"
5. Map the fourth layer: "What are my team's specific accountabilities that drive success at layer three?"
6. Communicate the cascade to the team — Not just what to do, but why it connects upward

What this fixes: Teams "wandering in the wilderness" — shipping work that doesn't connect to strategy because the context was never translated for them.

The core principle: Even with incomplete direction from above, a Director's job is to fill the gap downward. Waiting for perfect clarity is a PM habit. Creating imperfect-but-useful clarity is a Director skill.

Application

Using This Framework as a PM (Pre-Transition)

1. Identify which transition zone you're weakest in — not to act on it yet, but to know what to observe
2. Use 1-on-1s with your manager to practice Zone 4: "How does my work connect to business strategy? What's the organizational context I'm not seeing?"
3. Watch for Hero Syndrome habits now: do you jump in to solve things that others could solve with your coaching?
4. Don't over-invest in Director thinking while you're still in a PM role. Serve your current scope with full commitment — director altitude will be available when the context requires it

Using This Framework as a Newly Promoted Director

1. First 30 days: Draw your new Altitude & Horizon map. Who are your new stakeholders? What does a quarter-to-annual planning horizon actually look like in this organization?
2. First 60 days: Identify your Hero Syndrome triggers. When do you feel the pull to jump in directly instead of coaching?
3. First 90 days: Run your first Cascading Context Map. Even if company strategy is unclear, make your best translation and share it with your team
4. Ongoing: When friction appears, name which transition zone it lives in. Diagnosis before prescription

Running a Cascading Context Map

Use when your team is unclear on what organizational strategy means for their work.

## Context Cascade

**Company Priority:** [What leadership said — in their words]
**Business Unit Translation:** [How your BU contributes to that priority]
**Product Portfolio Translation:** [How your products contribute to that]
**Team Accountabilities:** [What each team owns specifically]
**Why this matters:** [The so-what for your team — what changes, what stays the same]

One page is better than ten. The goal is clarity, not comprehensiveness.

Examples

See examples/sample.md for a full worked scenario with a completed Cascading Context Map and anti-pattern contrast.

Good: Director Creates Clarity from a Vague Company Priority

Situation: CEO announces at QBR: "We're doubling down on enterprise." Three PMs ask their Director: "What does that mean for our roadmaps?"

PM response (wrong altitude): "Let's add enterprise features to our sprint backlogs."

Director response (right altitude): Runs a Cascading Context Map. Translates: "Enterprise means larger deal sizes, longer sales cycles, and more integration requirements. For our portfolio: Product A owns the admin controls story, Product B owns the API documentation story, Product C owns the security certification story. Here's what changes in Q3 planning and what doesn't."

Why it works: Director didn't wait for more clarity. They created it from available signal.

Bad: Hero Syndrome in Action

Situation: A PM on the team is struggling with a difficult stakeholder relationship.

Director response (wrong): "Let me just talk to that stakeholder directly — I'll get it sorted out."

Director response (right): "Walk me through what you've tried. Let's figure out where it broke down and what you'll do differently."

Why it matters: The first response solves the problem and creates dependency. The second response grows the PM. Directors who rescue too often build teams that can't function without them.

Good: Shifting from Waiter to Operator

Situation: A high-performing PM insists on documenting requirements in a different format from the rest of the team because "my stakeholders prefer it."

Director response (wrong): "That's fine, she's our best PM — if it works for her team, let it go."

Director response (right): "Joe is crushing it individually. But when marketing tries to synthesize across all three PMs' work, they can't. Shared process isn't bureaucracy — it's what makes the system legible to everyone outside it."

Why it matters: Protecting high-performer exceptions creates invisible coordination costs. The Restaurant Operator's job is the system, not the star waiter.

Common Pitfalls

Pitfall 1: Altitude Theater

Symptom: Using strategy language ("portfolio," "ecosystem," "long-term vision") while still making sprint-level decisions

Consequence: You sound like a Director but function like a PM. Your team is confused about who's actually deciding and at what level.

Fix: If you're in the details, own it. If you're not, delegate it fully. Mixing altitude levels without signaling creates ambiguity that erodes team trust.

Pitfall 2: One-and-Done Context Cascade

Symptom: Running the Cascading Context Map once at annual planning, then never revisiting it

Consequence: Team aligns in Q1 and drifts as strategy evolves. By Q3, team work is decoupled from current priorities.

Fix: Revisit the cascade at major inflection points — quarterly planning, significant exec changes, pivots, or org restructuring.

Pitfall 3: Confusing Kindness with Leadership

Symptom: Shielding the team from hard decisions, over-explaining constraints you're holding, softening feedback into meaninglessness

Consequence: Team operates without accurate context; trust erodes when reality eventually lands without warning.

Fix: Be transparent about the "why" behind hard decisions. You don't need to share everything — but what you share should be honest and actionable.

Pitfall 4: Premature Director Thinking as a PM

Symptom: Spending PM years worried about portfolio strategy, organizational dynamics, and "thinking above your pay grade"

Consequence: You under-serve your current role. PMs who think like Directors often miss the customer-level signal their actual role requires.

Fix: Play your current role with full commitment. The transition will demand Director thinking soon enough — you'll be ready because you did your PM work well, not because you rehearsed the Director role prematurely.

References

Related Skills

• skills/director-readiness-advisor/SKILL.md — Interactive advisor that uses this framework to diagnose and coach your specific transition situation

Source Material

• The Product Porch, Episode 42: From Product Manager to Director: How to Make the Shift (Part 1) — Todd Blaquiere, Ryan Cantwell, Joe Ghali (January 2026)

External Frameworks

• Marty Cagan, Empowered — Organizational dynamics and role clarity in product leadership
• Julie Zhuo, The Making of a Manager — IC-to-manager transition with practical war stories
• Michael Watkins, The First 90 Days — Structured approach to leadership transitions

Purpose

Diagnose overall SaaS business health by analyzing growth, retention, unit economics, and capital efficiency metrics together. Use this to identify problems early, prioritize actions by urgency, and deliver a comprehensive health scorecard for board meetings, quarterly reviews, or fundraising preparation.

This is not a single-metric check—it's a holistic diagnostic that connects revenue, retention, economics, and efficiency to reveal systemic issues and opportunities.

Input

Works best with: Whatever business metrics you have access to — growth rate, NRR/GRR, CAC payback, burn multiple, runway. Partial data is fine; the diagnostic flags what's missing. Also useful: The occasion (board meeting, quarterly review, fundraise prep) and what you already suspect is wrong.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The diagnostic opens by asking your company stage and which metric families you can pull.

Example invocation: Run a health check: $8M ARR, 60% YoY growth, NRR 96%, CAC payback 21 months, 14 months runway — board meeting in 3 weeks.

Key Concepts

The Business Health Framework

A SaaS business is healthy when four dimensions work together:

1. Growth & Retention — Are you growing and keeping customers?

   • Revenue growth rate
   • NRR (Net Revenue Retention)
   • Churn rate
   • Quick Ratio

2. Unit Economics — Is the business model profitable at the customer level?

   • CAC (Customer Acquisition Cost)
   • LTV (Lifetime Value)
   • LTV:CAC ratio
   • Payback period
   • Gross margin

3. Capital Efficiency — Are you using cash efficiently?

   • Burn rate
   • Runway
   • Rule of 40
   • Magic Number

4. Strategic Position — Are you positioned for sustainable success?

   • Market positioning (below, at, above market pricing)
   • Competitive moat (network effects, data, brand)
   • Revenue concentration risk
   • Operating leverage

Stage-Specific Benchmarks

Early Stage (Pre-$10M ARR):

• Focus: Product-market fit, unit economics
• Growth: >50% YoY
• LTV:CAC: >3:1
• Gross Margin: >70%
• Runway: >12 months
• Acceptable: Negative margins, high burn (if unit economics work)

Growth Stage ($10M-$50M ARR):

• Focus: Scaling efficiently
• Growth: >40% YoY
• NRR: >100%
• Rule of 40: >40
• Magic Number: >0.75
• Acceptable: Moderate burn if growth is strong

Scale Stage ($50M+ ARR):

• Focus: Profitability, efficiency
• Growth: >25% YoY
• NRR: >110%
• Rule of 40: >40
• Profit Margin: >10%
• Required: Positive or near-positive cash flow

Red Flag Categories

Critical (Fix immediately):

• Runway <6 months
• LTV:CAC <1.5:1
• Churn accelerating cohort-over-cohort
• NRR <90%
• Magic Number <0.3

High Priority (Fix within quarter):

• Rule of 40 <25
• Payback >24 months
• Quick Ratio <2
• Gross margin <60%
• Revenue concentration >50% in top 10 customers

Medium Priority (Address within 6 months):

• NRR 90-100% (flat, not growing)
• Magic Number 0.3-0.5
• Operating leverage negative
• Churn rate stable but high (>5% monthly)

Anti-Patterns (What This Is NOT)

• Not a single metric: "Revenue is growing 50%, we're great!" (ignoring burn, churn, unit economics)
• Not stage-agnostic: Early-stage burn is acceptable; scale-stage burn is a problem
• Not static: Health is directional—are metrics improving or degrading?
• Not just numbers: Context matters (competitive pressure, market changes, team capacity)

When to Use This Framework

Use this when:

• Preparing for board meetings or investor updates
• Quarterly business reviews (QBR)
• Fundraising preparation (know your numbers)
• Annual planning (identify improvement areas)
• You suspect problems but can't pinpoint them
• New PM/exec joining and needs health assessment

Don't use this when:

• You're pre-revenue (focus on product-market fit first)
• You're in pure research mode (not enough data)
• You need tactical guidance (use specific skills: feature, channel, pricing)

Facilitation Source of Truth

Use workshop-facilitation as the default interaction protocol for this skill.

It defines:

• session heads-up + entry mode (Guided, Context dump, Best guess)
• one-question turns with plain-language prompts
• progress labels (for example, Context Qx/8 and Scoring Qx/5)
• interruption handling and pause/resume behavior
• numbered recommendations at decision points
• quick-select numbered response options for regular questions (include Other (specify) when useful)

This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.

Application

This interactive skill asks up to 4 adaptive questions, then delivers a comprehensive diagnostic with prioritized recommendations.

Step 0: Gather Context

Agent asks:

"Let's diagnose your business health. I'll need metrics across four dimensions: growth, retention, unit economics, and capital efficiency.

Company context:

• Stage: (Pre-$10M ARR, $10M-$50M ARR, $50M+ ARR)
• Business model: (PLG, sales-led, hybrid)
• Target market: (SMB, mid-market, enterprise, mixed)

Why this matters: Benchmarks vary by stage. Early-stage optimizes for growth; scale-stage optimizes for efficiency.

Please provide the following metrics. Use 'unknown' if you don't have a metric."

Step 1: Growth & Retention Metrics

Agent asks:

"Growth & Retention:

1. Revenue:

   • Current MRR or ARR: $___
   • Revenue growth rate: ___% (MoM or YoY)

2. Retention:

   • Monthly churn rate: ___%
   • NRR (Net Revenue Retention): ___%
   • Quick Ratio: ___ (or I can calculate it)

3. Expansion:

   • Expansion revenue as % of total MRR: ___%

4. Cohort trends:

   • Are recent cohorts retaining better or worse than older cohorts?
     1. Better (improving)
     2. Same (stable)
     3. Worse (degrading)
     4. Unknown"

Based on answers, agent evaluates:

• ✅ Healthy growth: Growth >40% YoY (growth stage) or >25% (scale stage)
• ✅ Healthy retention: NRR >100%, churn <5% monthly, Quick Ratio >2
• 🚨 Growth problems: Growth <20% YoY
• 🚨 Retention problems: NRR <100%, churn >5%, cohort degradation

Step 2: Unit Economics Metrics

Agent asks:

"Unit Economics:

1. Acquisition:

   • CAC (Customer Acquisition Cost): $___
   • Blended or by channel? (If by channel, what's your best channel CAC?)

2. Value:

   • LTV (Lifetime Value): $___
   • LTV:CAC ratio: ___ (or I can calculate it)
   • Payback period: ___ months (or I can calculate it)

3. Margins:

   • Gross margin: ___%
   • Contribution margin (if known): ___%

4. Trends:

   • Is CAC increasing, stable, or decreasing over time?
     1. Decreasing (improving efficiency)
     2. Stable
     3. Increasing (diminishing returns)
     4. Unknown"

Based on answers, agent evaluates:

• ✅ Healthy economics: LTV:CAC >3:1, payback <12 months, gross margin >70%
• ⚠️ Marginal economics: LTV:CAC 2-3:1, payback 12-18 months
• 🚨 Poor economics: LTV:CAC <2:1, payback >24 months, gross margin <60%

Step 3: Capital Efficiency Metrics

Agent asks:

"Capital Efficiency:

1. Cash:

   • Cash balance: $___
   • Monthly net burn rate: $___
   • Runway: ___ months (or I can calculate it)

2. Efficiency ratios:

   • Rule of 40: ___ (Growth % + Profit Margin %) (or I can calculate it)
   • Magic Number: ___ (S&M efficiency) (or I can calculate it)

3. Operating expenses:

   • S&M as % of revenue: ___%
   • R&D as % of revenue: ___%
   • Is OpEx growing faster than revenue?
     1. No (positive operating leverage)
     2. Yes (negative operating leverage)
     3. Unknown

4. Profitability:

   • Profit margin: ___%
   • Path to profitability: (already profitable, 6-12 months, 12-24 months, >24 months, unknown)"

Based on answers, agent evaluates:

• ✅ Healthy efficiency: Rule of 40 >40, magic number >0.75, runway >12 months
• ⚠️ Acceptable efficiency: Rule of 40 25-40, magic number 0.5-0.75, runway 6-12 months
• 🚨 Poor efficiency: Rule of 40 <25, magic number <0.5, runway <6 months

Step 4: Deliver Comprehensive Diagnostic

Agent synthesizes all metrics and delivers:

1. Overall Health Score — Healthy / Moderate / Concerning / Critical
2. Dimension Scores — Growth, Retention, Economics, Efficiency
3. Red Flags — Critical, High Priority, Medium Priority
4. Prioritized Recommendations — Top 3-5 actions with expected impact
5. Stage-Appropriate Benchmarks — How you compare to peers

Diagnostic Pattern 1: Healthy Business

When:

• Growth, retention, economics, and efficiency all meet stage-appropriate benchmarks
• No critical red flags
• Improving trends

Output:

"## ✅ Overall Health: Healthy

Your business shows strong fundamentals across all dimensions.

Health Scorecard

Key Strengths

1. [Specific strength 1]

   • Metric: [e.g., NRR 120%]
   • Why it matters: [Expanding within base without new logos]

2. [Specific strength 2]

   • Metric: [e.g., LTV:CAC 5:1]
   • Why it matters: [Sustainable unit economics support scaling]

3. [Specific strength 3]

   • Metric: [e.g., Rule of 40 = 65]
   • Why it matters: [Excellent balance of growth and efficiency]

Opportunities for Optimization

Even healthy businesses can improve. Here are your top opportunities:

1. [Opportunity 1]

• Current: [e.g., Magic Number 0.9]
• Opportunity: [Could scale S&M spend 2x and maintain efficiency]
• Impact: [+$___ MRR/month]

2. [Opportunity 2]

• Current: [e.g., Expansion revenue 15% of total]
• Opportunity: [Build upsell paths, target 25% expansion revenue]
• Impact: [NRR 110% → 120%]

3. [Opportunity 3]

• Current: [e.g., CAC $500, stable]
• Opportunity: [Improve conversion, reduce CAC to $400]
• Impact: [Faster payback, better LTV:CAC]

Recommended Actions (Next Quarter)

Priority 1: Scale what's working

• [e.g., Double content marketing budget (best channel)]
• Expected impact: [+___ customers/month, +$___ MRR]

Priority 2: Expand within base

• [e.g., Launch premium tier for 20% of customers]
• Expected impact: [NRR 110% → 115%]

Priority 3: Improve efficiency

• [e.g., Optimize paid acquisition (reduce CAC 10%)]
• Expected impact: [Payback 8mo → 7mo]

Monitor These Metrics

Weekly:

• NRR (should stay >___%)
• Churn rate (should stay <___%)
• Quick Ratio (should stay >___)

Monthly:

• Rule of 40 (should stay >___)
• Magic Number (should stay >___)
• LTV:CAC (should stay >___:1)

Quarterly:

• Cohort retention trends
• Revenue concentration risk
• Operating leverage

Benchmarks (Your Stage: [Growth/Scale])

You're performing at or above benchmarks across the board."

Diagnostic Pattern 2: Moderate Health (Fixable Issues)

When:

• Most metrics acceptable, but 1-2 dimensions have problems
• Medium-priority red flags
• Solvable with focus

Output:

"## ⚠️ Overall Health: Moderate (Fixable Issues)

Your business has good fundamentals but needs attention in [specific dimension].

Health Scorecard

Red Flags Identified

High Priority 🚨

1. [Specific red flag]
   • Metric: [e.g., NRR 95%]
   • Threshold: [Should be >100%]
   • Impact: [Base is contracting, not expanding]
   • Fix by: [End of quarter]

Medium Priority ⚠️

1. [Specific issue]
   • Metric: [e.g., Magic Number 0.6]
   • Threshold: [Should be >0.75]
   • Impact: [S&M spend moderately efficient, room for improvement]
   • Fix by: [6 months]

Root Cause Analysis

Primary Issue: [e.g., Retention & Expansion]

Symptoms:

• NRR 95% (should be >100%)
• Churn rate 5% monthly (should be <3%)
• Expansion revenue only 10% of MRR (should be 20-30%)

Diagnosis: [e.g., Customers are churning before they expand. Onboarding is weak, no clear upsell paths.]

Impact:

• Lost MRR: [Calculate churn impact]
• Missed expansion: [Calculate expansion opportunity]
• Total impact: [Combined revenue loss]

Prioritized Action Plan

Immediate (Next 30 days):

1. Fix [Primary Issue]

• Action: [Specific step, e.g., "Launch onboarding improvement program"]
• Owner: [PM, Customer Success]
• Target: [Reduce churn 5% → 4%]
• Impact: [Save $___K MRR/month]

Short-term (Next Quarter):

2. [Secondary Action]

• Action: [e.g., "Build premium tier for upsell"]
• Target: [NRR 95% → 105%]
• Impact: [+$___K expansion MRR]

3. [Tertiary Action]

• Action: [e.g., "Optimize S&M spend, improve magic number"]
• Target: [Magic Number 0.6 → 0.8]
• Impact: [More efficient growth]

What Success Looks Like (90 Days)

Target metrics:

• NRR: 95% → 105% (+10pp)
• Churn: 5% → 3.5% (-30%)
• Magic Number: 0.6 → 0.8 (+33%)

Impact:

• Monthly revenue saved from churn: +$___K
• Expansion revenue: +$___K
• More efficient S&M: [details]

If you hit these targets, you'll be in 'Healthy' territory.

Monitor Weekly

Must-track metrics:

• Churn rate (track to ensure it's decreasing)
• NRR (track to ensure it's improving)
• Customer feedback (are improvements working?)

Leading indicators:

• Onboarding completion rate
• Time-to-value
• Usage metrics (activation, engagement)

What Not to Do

Don't:

• Scale acquisition until retention is fixed (you'll just churn faster)
• Ignore expansion (it's easier than new acquisition)
• Wait too long (retention problems compound)"

Diagnostic Pattern 3: Concerning Health (Urgent Action Required)

When:

• Multiple critical red flags
• 2+ dimensions problematic
• Requires immediate intervention

Output:

"## 🚨 Overall Health: Concerning (Urgent Action Required)

Your business has multiple critical issues that need immediate attention.

Health Scorecard

Critical Red Flags 🚨

1. [Critical Issue 1 - e.g., Runway]

• Current: [6 months runway]
• Threshold: [<6 months = crisis]
• Impact: [Survival risk]
• Action: [Raise capital OR cut burn immediately]
• Timeline: [30 days]

2. [Critical Issue 2 - e.g., Unit Economics]

• Current: [LTV:CAC 1.2:1]
• Threshold: [<1.5:1 = unsustainable]
• Impact: [Losing money on every customer]
• Action: [Reduce CAC OR increase LTV]
• Timeline: [60 days]

3. [Critical Issue 3 - e.g., Cohort Degradation]

• Current: [Newer cohorts churning 2x faster than old]
• Threshold: [Degrading PMF]
• Impact: [Scaling makes problem worse]
• Action: [Stop scaling, fix retention]
• Timeline: [90 days]

Survival Plan (Next 90 Days)

Week 1-2: Triage

Immediate actions:

1. Extend runway (if <6 months)

   • Option A: Raise bridge round ($___K)
   • Option B: Cut burn by ___%
   • Option C: Combination
   • Decision by: [Date]

2. Stop scaling broken channels

   • Pause S&M spend on channels with LTV:CAC <2:1
   • Reallocate budget to [best-performing channel]

3. Assemble crisis team

   • Daily standups on key metrics
   • Weekly progress reviews

Month 1: Stop the Bleeding

Priority 1: Fix Unit Economics

• Current: LTV:CAC ___:1 (unsustainable)
• Actions:
  1. Reduce CAC: [Specific tactics]
  2. Increase LTV: [Improve retention, add expansion]
• Target: LTV:CAC >2:1 within 30 days

Priority 2: Improve Retention

• Current: Churn ___% (too high)
• Actions:
  1. Interview churned customers (identify top 3 reasons)
  2. Fix onboarding (reduce early churn)
  3. Proactive outreach to at-risk accounts
• Target: Reduce churn by 20% within 30 days

Month 2-3: Stabilize

Milestone 1: Positive Unit Economics

• LTV:CAC >2:1 ✅
• Payback <18 months ✅
• Gross margin >60% ✅

Milestone 2: Slowing Churn

• Churn decreasing month-over-month
• Cohort degradation stopped
• NRR improving toward 100%

Milestone 3: Runway Extended

• 12+ months runway (via fundraise or burn reduction)
• Clear path to next milestone

What Success Looks Like (Day 90)

Metrics:

• Runway: ___ months → 12+ months ✅
• LTV:CAC: ___:1 → >2:1 ✅
• Churn: ___% → reduced by 30% ✅
• NRR: ___% → improving toward 100%

Position:

• Out of crisis mode
• Stable foundation to rebuild growth
• Clear plan for next 6-12 months

What to Avoid

Don't:

• Try to grow your way out of this (fix unit economics first)
• Ignore the data (hope is not a strategy)
• Scale before you fix retention (accelerates failure)
• Wait until runway <3 months to fundraise (too late)

Do:

• Focus ruthlessly on retention and unit economics
• Cut costs to extend runway
• Be honest with board/investors about problems
• Move fast (you don't have time to waste)"

Diagnostic Pattern 4: Critical Health (Existential Crisis)

When:

• Runway <3 months OR
• Multiple critical failures (LTV:CAC <1:1, massive churn, no path to profitability)

Output:

"## 🚨🚨 Overall Health: Critical (Existential Crisis)

Your business is in survival mode. Immediate drastic action required.

[Similar structure to Pattern 3, but more urgent tone, shorter timelines, more drastic measures]

Immediate Actions (This Week):

1. Emergency board meeting
2. Fundraise immediately OR cut burn 50%+
3. Stop all non-essential spend
4. Fix top 1-2 critical issues (runway, unit economics)"

Examples

See examples/ folder. Mini examples below:

Example 1: Healthy Growth-Stage SaaS

Metrics:

• ARR: $20M, Growth: 60% YoY
• NRR: 115%, Churn: 2.5%
• LTV:CAC: 4:1, Payback: 10 months
• Rule of 40: 50, Runway: 18 months

Diagnosis: Healthy. Scale aggressively.

Example 2: Moderate Health (Retention Issue)

Metrics:

• ARR: $15M, Growth: 40% YoY
• NRR: 95%, Churn: 5%
• LTV:CAC: 3.5:1, Payback: 12 months
• Rule of 40: 38, Runway: 12 months

Diagnosis: Moderate. Fix retention before scaling further.

Example 3: Concerning (Multiple Issues)

Metrics:

• ARR: $8M, Growth: 25% YoY (slowing)
• NRR: 88%, Churn: 7% (increasing)
• LTV:CAC: 1.8:1, Payback: 20 months
• Rule of 40: 15, Runway: 8 months

Diagnosis: Concerning. Urgent action on retention and unit economics required.

Common Pitfalls

Pitfall 1: Celebrating Single Metrics

Symptom: "Revenue growing 50%!" (ignoring burn, churn, unit economics)

Consequence: Unsustainable growth. Scaling broken model.

Fix: Look at all four dimensions together.

Pitfall 2: Ignoring Stage-Specific Benchmarks

Symptom: "We're not profitable yet, is that bad?" (early-stage company)

Consequence: Misplaced worry. Early-stage should optimize for growth and unit economics, not profitability.

Fix: Use stage-appropriate benchmarks.

Pitfall 3: Focusing on Lagging Indicators Only

Symptom: "Churn is 5%, let's watch it"

Consequence: By the time lagging indicators (churn, NRR) show problems, it's late.

Fix: Track leading indicators (usage, engagement, onboarding completion).

Pitfall 4: Not Acting on Red Flags

Symptom: "NRR <100% for 3 quarters, but we'll fix it eventually"

Consequence: Problems compound. Becomes crisis.

Fix: Set clear timelines. If metric doesn't improve in X time, escalate.

Pitfall 5: Trying to Fix Everything at Once

Symptom: "Let's improve growth, retention, CAC, and efficiency simultaneously"

Consequence: Resources spread thin. Nothing improves.

Fix: Prioritize top 1-3 issues. Fix sequentially.

References

Related Skills

• saas-revenue-growth-metrics — Detailed growth and retention metrics
• saas-economics-efficiency-metrics — Detailed unit economics and capital efficiency
• finance-metrics-quickref — Fast lookup for all metrics and benchmarks
• feature-investment-advisor — Uses health diagnostic to inform feature priorities
• acquisition-channel-advisor — Uses health diagnostic to inform channel priorities
• finance-based-pricing-advisor — Uses health diagnostic to inform pricing decisions

External Frameworks

• Bessemer Venture Partners: "SaaS Metrics 2.0" — Comprehensive benchmarks
• David Skok: "SaaS Metrics" — Unit economics benchmarks
• OpenView Partners: SaaS benchmarking reports
• Battery Ventures: "State of SaaS" annual report

Provenance

• Adapted from research/finance/Finance_QuickRef.md (Red flags table)
• Decision frameworks from research/finance/Finance_For_PMs.Putting_It_Together_Synthesis.md
• Benchmarks from research/finance/Finance for Product Managers.md

Purpose

Research engine that builds deep, structured understanding of companies, industries, and competitor sets. Produces a stable output format that you can hand off to other skills and agents to generate battlecards, SWOT analyses, positioning statements, PESTEL assessments, market sizing, and workshop content.

This is not a generic encyclopedia lookup. Every section pushes toward commercial understanding, product implications, and actionable intelligence. The output is a research primitive — structured data other skills consume — not a final deliverable.

Input

Works best with: The research target: a company, an industry, or a set of competitors. Also useful: The downstream use (battlecard, SWOT, positioning, market sizing) so the research emphasizes the right lenses, plus any constraints on depth or recency.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The workflow opens by asking for the target and what the intel will feed.

Example invocation: Research Figma and its top 3 competitors — output feeds a positioning workshop next week.

Key Concepts

Four Entry Points

The skill auto-detects entry point from the user's input. If ambiguous, ask one clarifying question: "Is this about a specific company, an industry, or a set of competitors?"

Single Company — User names a company (e.g., "Parker Hannifin," "Novo Nordisk"). Produce the full 11-section output for that company.

Industry/Sector — User names an industry, sector, or niche (e.g., "clinical data management," "embedded finance," "upstream oil and gas"). Establish broad industry context, narrow into the segment, and connect findings to PM implications. Use the same 11-section structure adapted for sector-level analysis.

Named Competitor Set — User names 2-5 companies (e.g., "Compare Emerson, Honeywell, and Parker Hannifin"). Produce individual 11-section outputs for each company, then add a Section 12: Cross-Company Comparison that synthesizes across the set.

Discover Competitors — User names a company plus the word "competitors" (e.g., "productside.com competitors" or "Parker Hannifin competitors"). The skill:

1. Researches the named company first — enough to understand what it does, who it serves, and what market it plays in (a lightweight pass through Lenses 1-4)
2. Identifies 3-5 likely competitors based on that research, citing why each is a competitor (direct, adjacent, substitute, or emerging disruptor)
3. Presents the list for confirmation: "Based on my research, [Company]'s closest competitors appear to be [A, B, C, D, E]. Want me to run the full competitor set on these, or adjust the list first?"
4. Once confirmed, proceeds with the Named Competitor Set flow — full 11-section output for each company plus Section 12 cross-company comparison

The user can also provide a URL instead of a company name (e.g., "productside.com"). The skill should resolve the URL to the company, research accordingly, and proceed.

Seven Research Lenses

These lenses structure all analysis. Apply every lens to every entry point.

Lens 1 — Financial Landscape and Business Outcomes How the entity makes money. Major revenue streams and cost drivers, margin pressures, growth levers, retention and expansion dynamics, capital intensity, seasonal or cyclical patterns, major risks to performance.

Lens 2 — Market Offer and Business Model How the entity creates and captures value. Target markets, buyers, users, influencers, administrators, and blockers. How segments differ. Multi-sided or multi-stakeholder dynamics.

Lens 3 — Product Portfolio and Product Outcomes Major offers, product families, services, platforms, channels. Bundled solutions, ecosystem plays. Digital versus human-assisted components. Legacy versus emerging offers. Distinction between business line, offer, product, feature set, service layer, and enabling platform.

Lens 4 — Competitive Dynamics Direct competitors, adjacent competitors, substitutes, emerging disruptors. Where differentiation is won or lost.

Lens 5 — Rising Trends and Strategic Concerns Market trends, regulatory forces, technology shifts (especially AI and automation), operational constraints, buyer expectation changes, threats from consolidation or commoditization.

Lens 6 — How Product Management Works Here Product-led vs sales-led vs service-led behavior. Centralized vs federated product structures. Platform vs solution orientation. Roadmap and innovation posture. Compliance or governance overhead. Discovery maturity, data maturity, experimentation maturity, AI maturity. Cross-functional friction. Label inferences clearly.

Lens 7 — Strategic Signals Three signal types — always check all three:

• Patent activity: Recent filings and grants via patent databases. Technology domains, R&D clusters, gaps between patent activity and public product narrative.
• Hiring signals: Roles open in volume, skills and tools in job descriptions, seniority being hired, language that reveals product culture (discovery, outcomes, AI-native, regulatory, experimentation).
• Leadership changes: C-suite and product leadership arrivals or departures in the last 12-18 months. Origin of new leaders (platform companies, consulting, competitors). New CPO, CTO, or CDO roles created, eliminated, or restructured. Board-level changes.

Key Distinctions to Maintain

Always be disciplined about these — collapsing them produces shallow analysis:

• market vs segment
• buyer vs user
• product vs service
• business outcome vs product outcome vs output
• strategy vs tactics
• discovery vs delivery
• platform vs application
• signal vs assumption
• revenue growth vs market share growth vs customer lifetime value improvement vs cost reduction

Tensions Worth Surfacing

Highlight conflicts and tradeoffs wherever they appear:

• growth vs compliance
• scale vs customization
• digital self-service vs high-touch service
• standardization vs domain-specific workflow
• innovation vs legacy burden
• AI ambition vs governance reality
• customer value vs internal efficiency
• short-term revenue vs long-term platform investment

Why This Works

• Web-grounded: Uses live search, not training-data recall — output includes citations
• PM-native: Every section connects to product management implications, not just business facts
• Composable: Stable output format that downstream skills can parse and consume
• Repeatable: Same input next quarter produces fresh intel — the delta is the story
• Signal-driven: Strategic signals (patents, hiring, leadership) are often the most honest data available — they reveal what a company is actually doing, not what it says it's doing

Anti-Patterns

• Not a Wikipedia summary: Push past "what they do" to "what this means for product decisions"
• Not financial analysis: Focus is product strategy and commercial dynamics, not valuation or stock picks
• Not a prompt generator: The output is actual research with citations, not prompts for a future session
• Not a one-time exercise: Design for quarterly refresh — run it again, compare the delta

Research Expectations

Use web search actively. This skill requires live data gathering, not recall from training data. Search for and cite:

• Investor relations materials, annual reports, earnings transcripts
• Company product pages and official strategy pages
• Regulatory disclosures and filings
• Patent databases (Google Patents, USPTO)
• Company careers pages and job aggregators (LinkedIn, Indeed, Glassdoor)
• Executive appointment announcements and leadership change coverage
• Industry analysts (Gartner, Forrester, IDC) and reputable news coverage

Cite sources. Every factual claim should include a source. Separate fact from inference. When you're inferring — especially on Lens 6 (PM culture) and Lens 7 (strategic signals) — label it clearly: "Inference based on [evidence]."

Recency matters. Prioritize sources from the last 12-24 months. Flag anything older.

Application

Step 1: Detect Entry Point

Determine from user input:

• Single company → proceed to Step 2 with one entity
• Industry/sector → proceed to Step 2, adapt sections for sector-level analysis
• Named competitor set (2-5 companies listed) → proceed to Step 2 for each company, then Step 3
• Discover competitors (one company + "competitors") → proceed to Step 1b, then Step 2 for each, then Step 3
• URL provided (e.g., "productside.com") → resolve to company name, then detect entry point from any additional context

If ambiguous, ask one question: "Is this about a specific company, an industry, or a set of competitors?"

If the user provides additional context (e.g., "I'm preparing for a client engagement with them" or "we compete with them in the SMB segment"), use that context to weight which lenses get deeper treatment.

Step 1b: Discover Competitors (when entry point is "discover competitors")

1. Research the named company using web search. Do a lightweight pass through Lenses 1-4 — enough to understand what the company does, who it serves, what market it plays in, and how it creates value.

2. Identify 3-5 likely competitors based on that research. For each, state:

   • Company name
   • Why it's a competitor (direct, adjacent, substitute, or emerging disruptor)
   • One-sentence description of how it competes

3. Present the list for confirmation:

   "Based on my research, [Company] is [brief description — what it does and who it serves].

   Its closest competitors appear to be:

   1. [Competitor A] — [relationship: direct/adjacent/substitute/disruptor]. [Why.]
   2. [Competitor B] — [relationship]. [Why.]
   3. [Competitor C] — [relationship]. [Why.]
   4. [Competitor D] — [relationship]. [Why.]
   5. [Competitor E] — [relationship]. [Why.]

   Want me to run the full competitor set analysis on these? You can also add, remove, or swap any before I proceed."

4. Once confirmed, proceed to Step 2 for each company (including the original), then Step 3 (cross-company comparison).

Step 2: Research and Produce Output

Use web search to gather data across all seven lenses. Produce the following 11-section output:

## 1. What This Entity Is
[Business definition, founding, market position, scale. What makes it distinct.]

## 2. How It Makes Money
[Revenue streams, cost structure, margin dynamics, financial logic.
Seasonal or cyclical patterns. Growth levers and risks.]

## 3. Who It Serves
[Buyers, users, influencers, administrators, blockers.
Segment differences. Multi-stakeholder complexity.]

## 4. What It Sells or Delivers
[Core value propositions. Key offers in plain language.
How the offer creates value for the customer.]

## 5. Key Product Lines or Offers
[Mapped by product family, platform, service, channel.
Digital vs human-assisted. Legacy vs emerging.
Distinguish: business line, offer, product, feature set,
service layer, enabling platform.]

## 6. Business and Market Pressures
[Competitive forces, regulatory pressure, technology shifts,
operational constraints. Name the tensions.]

## 7. Competitors and Alternatives
[Direct competitors, adjacent competitors, substitutes,
emerging disruptors. Where differentiation is won or lost.]

## 8. Important Trends and Risks
[Macro forces, buyer expectation shifts, AI and automation impact,
consolidation or commoditization threats.]

## 9. Strategic Signals
[Patent activity: recent filings, technology domains, R&D bets.
Hiring signals: volume roles, skills language, seniority patterns.
Leadership changes: arrivals, departures, origins, new roles created.
Include sources for each signal.]

## 10. What This Means for Product Management
[PM implications: org dynamics, discovery maturity, delivery model,
cross-functional friction, AI readiness. Product-led vs sales-led.
Likely PM challenges. Domain-specific skills PMs would need.
Label inferences.]

## 11. Sources and Confidence
[List all sources used, organized by section.
Flag assumptions and inferences explicitly.
Note any sections where data was thin or unavailable.]

Quality checks for every section:

• Does it push past description to implication?
• Does it name specific tensions, not just facts?
• Are sources cited?
• Is inference labeled?

Step 3: Cross-Company Comparison (Competitor Set Only)

When the entry point is a competitor set, produce individual Section 1-11 outputs for each company, then add:

## 12. Cross-Company Comparison

### Where They're Betting Differently
[Patent clusters, hiring patterns, leadership hires that diverge.
Which companies are investing in AI, which in services,
which in platform plays.]

### Where They're Converging
[Same platform moves, same market pivots, same talent profiles.
When everyone zigs together, that's table stakes — not differentiation.]

### Gaps and White Space
[What none of them are covering. Segments underserved.
Capabilities nobody is building. Buyer needs unaddressed.]

### Tensions That Play Out Differently
[e.g., Company A chose scale over customization;
Company B chose the opposite. Who's winning, and for whom?]

### PM Implications Across the Set
[What a PM at each company would face differently.
Which org is better set up for discovery?
Which is most constrained by legacy?]

Step 4: Handoff Menu

After producing the output, offer the user a handoff menu. Each option names what gets built and which skill or agent consumes the research:

"Your research is ready. What do you want to build from it?

1. Competitive battlecard — I'll structure a head-to-head comparison for your sales or strategy team
2. SWOT analysis — I'll run strengths, weaknesses, opportunities, and threats using the research as input
3. Positioning statement — Use positioning-statement skill with this company/market context loaded
4. PESTEL assessment — Use pestel-analysis skill with the trends and pressures from Sections 6 and 8
5. Market sizing (TAM/SAM/SOM) — Use tam-sam-som-calculator skill with the market and segment data from Sections 2-3
6. Research prompts for deeper investigation — Generate 3-5 targeted research prompts per lens for a follow-up session
7. PM briefing memo — Condense the 11 sections into a 1-page executive summary for a PM audience
8. Workshop discussion guide — Extract teachable tensions and case study angles for training use

Select a number, combine them (e.g., '1 and 4'), or describe what you need."

Rerun Pattern

When the user reruns the skill on a previously researched entity:

• Focus web search on recent developments (last 90 days)
• Lead the output with a "What's Changed" summary before the full 11 sections
• In Section 9 (Strategic Signals), highlight delta from prior run: new hires, new patents, leadership moves
• Flag sections where nothing material changed: "No significant change since [prior date]"

The user does not need to say "refresh" — if the agent has prior output in context, it should default to delta-first reporting.

Examples

Example: Single Company — Parker Hannifin

Trigger: "Run company-intel on Parker Hannifin"

Entry point: Single Company

Section 1 excerpt: Parker Hannifin is a Fortune 250 diversified industrial manufacturer headquartered in Cleveland, Ohio, specializing in motion and control technologies. Founded in 1917, it operates across two segments: Diversified Industrial (~85% of revenue) and Aerospace Systems (~15%). The 2023 acquisition of Meggitt for $8.8B significantly expanded its aerospace portfolio.

Section 9 excerpt:

• Patents: Clustering in electro-hydraulic controls and hydrogen fuel cell components. R&D investment in electrification outpacing public product announcements — signals a bet on industrial decarbonization. (Source: Google Patents, 2024-2025 filings)
• Hiring: Volume hiring for "digital twin" engineers and IoT platform architects in the Diversified Industrial segment. Job descriptions reference AWS IoT and Azure Digital Twins. (Source: LinkedIn, Indeed — June 2026)
• Leadership: New VP of Digital Transformation hired from Rockwell Automation (2025). New Group President for Engineered Materials from Honeywell (2024). Pattern: importing talent from platform-oriented industrials. (Source: company press releases)

Section 10 excerpt: PMs at Parker face the classic industrial tension: long product lifecycles (10-20 years) vs. pressure to digitize and create recurring-revenue service layers. Product management is historically engineering-led, not customer-led. Discovery is constrained by the fact that customers (OEMs, utilities, defense contractors) have long procurement cycles and low tolerance for experimentation. The hiring signals suggest a push toward platform thinking, but the org structure (segment-based P&Ls) creates incentives to optimize locally rather than build horizontal platforms. Inference: the digital twin hiring is likely ahead of organizational readiness to consume it.

Example: Competitor Set — Industrial Motion Control

Trigger: "Compare Parker Hannifin, Emerson Electric, and Honeywell on company-intel"

Entry point: Competitor Set (3 companies)

Section 12 excerpt (Cross-Company Comparison):

Where They're Betting Differently:

• Parker is investing heavily in electrification and hydrogen (patent evidence). Emerson is betting on software-defined automation (Aspen Technology acquisition). Honeywell is splitting into three companies and doubling down on aerospace autonomy.

Gaps and White Space:

• None of the three have a credible PLG motion for their digital products — all rely on enterprise sales. A startup that cracks self-serve industrial IoT tooling could undercut all three on adoption speed.

PM Implications Across the Set:

• Parker PM = engineer-first, platform-curious but segment-siloed
• Emerson PM = software-led post-AspenTech, navigating legacy OT culture
• Honeywell PM = post-split identity crisis, aerospace PMs and industrial PMs now in different companies

Anti-Pattern Example

Weak: "Parker Hannifin makes industrial equipment and has strong financials."

Strong: Identifies the tension between Parker's motion-and-control platform business (recurring revenue, long service cycles) and its push into intelligent manufacturing and IIoT — and explains why that tension creates specific PM challenges around build-vs-partner decisions, aftermarket monetization, and the pace of digital product adoption in asset-intensive industries.

Common Pitfalls

Pitfall 1: Surface-Level Description Without Implications

Symptom: Summary reads like a Wikipedia article or press release. Consequence: No actionable intelligence. Downstream skills get nothing useful. Fix: Push every section toward "what does this mean for product decisions?" If a fact doesn't connect to a tension, tradeoff, or PM implication, it's not pulling its weight.

Pitfall 2: Skipping Strategic Signals

Symptom: Analysis draws only from press releases and About pages. Sections 1-8 are solid; Section 9 is empty or generic. Consequence: You're seeing what the company says it's doing, not what it's actually doing. Patents, hiring, and leadership changes are often the most honest signals available. Fix: Always search patents, hiring, and leadership as a required step — even if the results are thin. "No significant patent activity found" is a signal too.

Pitfall 3: Confusing Outputs With Outcomes

Symptom: Listing features or products without explaining what results they produce for customers or the business. Consequence: Section 5 becomes a product catalog instead of strategic intelligence. Fix: For every offer, answer: what problem does it solve, for whom, what outcome does it improve, and what behavioral change does it create?

Pitfall 4: No Citations

Symptom: "The CEO said the company is focused on AI." No source, no date, no context. Consequence: Unverifiable claims. Downstream consumers can't trust the research. Fix: Cite source and date. "CEO Jane Doe stated X in Q1 2026 earnings call (Source: Seeking Alpha transcript, Feb 2026)."

Pitfall 5: Treating All Industries as Identical

Symptom: Generic PM frameworks applied without domain calibration. "They should do more discovery" without acknowledging that discovery in defense contracting looks nothing like discovery in consumer SaaS. Consequence: Section 10 is useless to anyone who actually works in the domain. Fix: Identify what makes PM different in this specific domain — regulatory overhead, buyer/user separation, capital intensity, sales cycle length, service dependency.

Pitfall 6: One-Time Exercise, Never Refreshed

Symptom: Intel gathered once, never updated. Decisions made on 18-month-old hiring signals. Consequence: Stale intelligence is worse than no intelligence — it creates false confidence. Fix: Set a rerun cadence (quarterly for active competitors, annually for industry context). When rerunning, lead with "What's Changed."

Downstream Composition Guide

This section is for other skill authors and agent builders who want to consume company-intel output.

What This Skill Produces

A structured markdown document with 11 numbered sections (12 for competitor sets). Each section has a stable heading and defined content type:

How to Reference This Skill

In your skill's References section:

- **[company-intel](../company-intel/SKILL.md)** (Workflow) — Run first to generate structured company/industry research; this skill consumes Sections [X, Y, Z] as input

Passing Output to Downstream Skills

When handing off to a downstream skill, pass the relevant sections explicitly:

• Battlecard → Sections 4, 5, 7, 9, 12
• SWOT → Sections 2, 6, 7, 8, 9
• Positioning → Sections 3, 4, 7
• PESTEL → Sections 6, 8, 9
• TAM/SAM/SOM → Sections 2, 3, 5
• Business health → Sections 2, 5, 8
• PM briefing → Sections 1, 5, 9, 10
• Workshop guide → Sections 6, 9, 10 (tensions and PM implications)

References

Related Skills

• company-research (Component) — Lighter-weight company profile focused on executive quotes and product strategy; company-intel is deeper and broader, producing structured output for downstream consumption
• pestel-analysis (Component) — Deep PESTEL template; consumes Sections 6 and 8 from company-intel
• derisk-measurement-advisor (Interactive) — Risk scan that benefits from Sections 6 and 8 as context input
• tam-sam-som-calculator (Component) — Market sizing; consumes Sections 2, 3, 5
• business-health-diagnostic (Interactive) — SaaS health diagnosis; consumes Section 2
• positioning-statement (Component) — Positioning; consumes Sections 3, 4, 7
• acquisition-channel-advisor (Interactive) — Go-to-market analysis; consumes Sections 2, 3, 7
• proto-persona (Component) — Persona creation; consumes Section 3

Research Sources

• Company investor relations pages — Annual reports, earnings transcripts, forward-looking statements
• Patent databases — Google Patents, USPTO
• Job aggregators — LinkedIn, Indeed, Glassdoor (hiring signals)
• Industry analysts — Gartner, Forrester, IDC
• Regulatory filings — SEC (US), Companies House (UK), relevant sector regulators
• News coverage — Reuters, Bloomberg, industry-specific publications

Provenance

• Adapted from Dean Peters' company-profile-executive-insights-research prompt and TAM-SAM-SOM prompt generator in the product-manager-prompts repo
• Incorporates the Seven Research Lenses framework developed for Productside enterprise PM training

Purpose

Create a comprehensive company profile that extracts executive insights, product strategy, transformation initiatives, and organizational dynamics from publicly available sources. Use this to understand competitive landscape, evaluate partnership opportunities, benchmark best practices, prepare for interviews, or inform market entry decisions by understanding how successful companies think about product management and strategy.

This is not surface-level research—it's strategic intelligence gathering focused on product management perspectives and executive vision.

Input

Works best with: The company to profile. Also useful: Why you're researching it (interview prep, competitive analysis, partnership, market entry) — this changes what the brief emphasizes.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The skill asks for the company and the purpose before drafting the brief.

Example invocation: Build a company research brief on Datadog — I interview with their platform PM team Friday.

Key Concepts

The Executive Insights Framework

This framework synthesizes company intelligence across multiple dimensions:

Core Components:

1. Company Overview: Basic info, history, industry context
2. Executive Quotes: Strategic vision from CEO, COO, VP Product, Group PM
3. Product Insights: Strategy, recent launches, innovation focus
4. Transformation Strategies: Digital, AI, Agile transformations
5. Organizational Impact: How PM influences strategy, cross-functional collaboration
6. Future Roadmap: Upcoming initiatives and anticipated challenges
7. Product-Led Growth (PLG): PLG strategies, data-driven decisions

Why This Works

• Executive perspective: Captures leadership thinking, not just marketing copy
• Product-centric: Focuses on PM-relevant insights (strategy, process, culture)
• Multi-source: Synthesizes interviews, earnings calls, blog posts, case studies
• Strategic intelligence: Informs competitive positioning, partnership evaluation, or interview prep

Anti-Patterns (What This Is NOT)

• Not financial analysis: Focus is product strategy, not valuation or stock performance
• Not SWOT analysis: This documents their perspective, not strengths/weaknesses assessment
• Not surface scraping: Go deeper than "About Us" pages—find executive interviews, product blogs, earnings transcripts

When to Use This

• Competitive analysis (understanding how competitors approach PM)
• Partnership evaluation (assessing cultural fit and strategic direction)
• Interview preparation (understanding company culture, product philosophy)
• Benchmarking best practices (learning from successful companies)
• Market entry decisions (understanding how incumbents operate)

When NOT to Use This

• For internal analysis (this is external research)
• When primary sources are unavailable (executives haven't spoken publicly)
• As a substitute for customer research (this is company perspective, not customer perspective)

Application

Use template.md for the full fill-in structure.

Step 1: Define Research Scope

Clarify what you're researching and why:

## Research Objective
- **Company Name:** [e.g., "Stripe"]
- **Research Purpose:** [e.g., "Understand payment platform product strategy for competitive positioning"]
- **Key Questions:**
  - [Question 1: e.g., "How does Stripe think about platform extensibility?"]
  - [Question 2: e.g., "What's their approach to developer experience?"]
  - [Question 3: e.g., "How do they prioritize roadmap vs. custom enterprise requests?"]

Step 2: Gather Company Overview

Document basic company information:

### Company Overview

**Basic Information:**
- **Name:** [Official company name]
- **Headquarters:** [Location]
- **Industry:** [Primary industries, e.g., "Fintech, Payment Processing, Developer Tools"]
- **Founded:** [Year]
- **Size:** [Employees, revenue if public, funding if private]

**Brief History:**
- [Key milestones that shaped current market position]
- [Example: "2010: Founded by Patrick and John Collison. 2011: Launched 7-line integration. 2018: Launched Stripe Atlas. 2021: $95B valuation."]

Sources to check:

• Company website (About, Press, Blog)
• LinkedIn company page
• Crunchbase / PitchBook (for funding/valuation)
• Wikipedia (for history)

Step 3: Extract Executive Quotes on Strategic Vision

Find recent quotes from key executives:

### Executive Quotes on Strategic Vision

**Quote from the CEO:**
- "[Recent quote discussing long-term vision and market approach]"
- **Source:** [Link to interview, earnings call, blog post, conference talk]
- **Date:** [When the quote was made]
- **Context:** [Brief explanation of what prompted this quote]

**Quote from the COO:**
- "[Recent quote focusing on operational strategies and challenges]"
- **Source:** [Link]
- **Date:** [When]

**Quote from the VP of Product Management:**
- "[Recent quote detailing product strategy and innovation focus]"
- **Source:** [Link]
- **Date:** [When]

**Quote from the Group Product Manager:**
- "[Recent quote discussing specific product initiatives and customer engagement]"
- **Source:** [Link]
- **Date:** [When]

Sources to check:

• Earnings call transcripts (if public)
• Podcast interviews (e.g., Lenny's Podcast, Masters of Scale, How I Built This)
• Conference talks (YouTube, company blog)
• Blog posts by executives
• LinkedIn posts
• Industry publications (TechCrunch, The Verge, etc.)

Quality checks:

• Recent: Prioritize quotes from the last 12-24 months
• Substantive: Look for strategy/philosophy, not generic PR statements
• Attributed: Always cite source and date

Step 4: Document Product Insights

Synthesize product strategy and recent launches:

### Detailed Product Insights

**Product Strategy Overview:**
- [Describe overall product strategy, emphasizing integration of market needs with technological capabilities]
- [Example: "Stripe's product strategy centers on developer experience: reduce integration complexity, provide powerful primitives, enable rapid experimentation"]

**Recent Product Launches and Innovations:**
1. **[Product/Feature 1]** - [Description and market impact]
   - [Example: "Stripe Tax (2021): Automated sales tax calculation. Removed compliance barrier for global expansion."]
2. **[Product/Feature 2]** - [Description and impact]
3. **[Product/Feature 3]** - [Description and impact]

**Product Philosophy:**
- [Key principles that guide product decisions]
- [Example: "Start with developer needs, not enterprise sales. Build for 10x scale before you need it. Default to public APIs."]

Sources to check:

• Product blog or changelog
• Product Hunt launches
• Release notes
• Product team blog posts or case studies

Step 5: Identify Transformation Strategies

Document how the company is evolving:

### Transformation Strategies and Initiatives

**Digital Transformation:**
- [Describe approach to digital transformation, emphasizing integration of cutting-edge technology with existing processes]
- [Example: "Migrated from monolith to microservices architecture (2019-2022). Enabled 10x faster feature deployment."]

**AI Transformation:**
- [Explain how AI is incorporated into core processes, product offerings, and market positioning]
- [Example: "Launched Radar for fraud detection (ML-powered). Reduced false positives by 40%, processing $640B annually."]

**Agile Transformation:**
- [Detail adoption of Agile methodologies, highlighting improvements in collaboration, project management, product delivery]
- [Example: "Adopted Shape Up methodology (6-week cycles, no sprints). Improved focus, reduced meeting overhead."]

Sources to check:

• Engineering blog
• Case studies or white papers
• Conference talks by engineering/product leaders
• LinkedIn posts about process changes

Step 6: Understand Organizational Impact of Product Management

Document how PM functions within the organization:

### Organizational Impact of Product Management

**Role of Product Management in Strategic Decisions:**
- [Discuss how PM influences strategic decisions]
- [Example: "PMs own P&L for their product area. Directly influence company roadmap through quarterly planning process. CEO reviews roadmap with PM leads, not just VPs."]

**Cross-Functional Collaboration:**
- [Outline collaboration between PM and other departments]
- [Example: "PMs co-located with engineering (not in separate 'product' org). Weekly design reviews with Design VP. Monthly GTM sync with Sales/Marketing."]

**PM Career Paths:**
- [If available, describe how PMs grow and advance]
- [Example: "IC track: PM → Senior PM → Staff PM → Principal PM. Manager track: PM → Group PM → Director → VP."]

Sources to check:

• PM job postings (describe role, responsibilities, team structure)
• LinkedIn profiles (track PM career progression)
• PM blog posts or interviews
• Glassdoor reviews (internal culture insights)

Step 7: Analyze Future Roadmap and Challenges

Identify where the company is headed:

### Future Product Roadmap and Challenges

**Upcoming Product Initiatives:**
- [Detail planned initiatives and alignment with strategic goals]
- [Example: "Expanding into embedded finance (Stripe Capital, Stripe Treasury). Goal: Become financial infrastructure for the internet, not just payments."]

**Anticipated Market Challenges:**
- [Identify potential challenges and PM team plans to address them]
- [Example: "Challenge: Increasing competition from Square, PayPal. Response: Double down on developer experience, global expansion (70+ countries)."]

**Competitive Threats:**
- [Document acknowledged or observed competitive pressures]

Sources to check:

• Earnings calls (forward-looking statements)
• Analyst reports
• Industry news (funding rounds by competitors, market shifts)

Step 8: Document Product-Led Growth Insights

If applicable, capture PLG strategies:

### Product-Led Growth Insights

**Implementation of PLG Strategies:**
- [Describe how the company employs PLG to enhance customer acquisition, retention, expansion]
- [Example: "Self-serve onboarding: 7-line code integration. No sales calls required for <$1M ARR. 90% of customers start with free tier."]

**Data-Driven Product Decisions:**
- [Explain role of data analytics in shaping product decisions and driving growth]
- [Example: "Instrumented every API call. PMs have real-time dashboards. Feature adoption tracked within 24 hours of launch."]

Sources to check:

• Product analytics blog posts
• Growth team blog posts
• Case studies on activation, retention, expansion

Step 9: Synthesize Key Takeaways

Summarize the most important insights:

### Key Takeaways

**Strategic Principles:**
1. **[Principle 1]** - [What you learned about their approach]
2. **[Principle 2]** - [What you learned]
3. **[Principle 3]** - [What you learned]

**Product Management Lessons:**
1. **[Lesson 1]** - [Applicable insight for your context]
2. **[Lesson 2]** - [Applicable insight]
3. **[Lesson 3]** - [Applicable insight]

**Questions for Further Research:**
- [Unanswered question 1]
- [Unanswered question 2]

Examples

See examples/sample.md for a full company research example.

Mini example excerpt:

**Company Name:** Stripe
**Research Purpose:** Understand payment platform product strategy
**Key Questions:** Developer experience? Platform extensibility?

Common Pitfalls

Pitfall 1: Surface-Level Research

Symptom: "Stripe is a payments company. They process payments."

Consequence: No strategic insights.

Fix: Go deeper—find executive interviews, engineering blogs, product philosophy posts.

Pitfall 2: No Source Citations

Symptom: "The CEO said the company is focused on innovation"

Consequence: Unverifiable, low credibility.

Fix: Always cite source and date: "The CEO said X (Source: Lenny's Podcast, Episode 185, Sept 2023)."

Pitfall 3: Mixing Opinion with Facts

Symptom: "Stripe's product strategy is great because they focus on developers"

Consequence: Analysis, not research.

Fix: Document what they do, not whether it's "good." Save analysis for "Key Takeaways."

Pitfall 4: Outdated Information

Symptom: Using 5-year-old quotes or strategies

Consequence: Irrelevant insights (company strategies evolve).

Fix: Prioritize sources from the last 12-24 months.

Pitfall 5: Ignoring Negative Signals

Symptom: Only documenting successes, ignoring challenges or failures

Consequence: Incomplete picture.

Fix: Include "Anticipated Market Challenges" and competitive threats.

References

Related Skills

• skills/positioning-statement/SKILL.md — Use company research to understand competitive positioning
• skills/pestel-analysis/SKILL.md — Company research informs market context
• skills/proto-persona/SKILL.md — Executive quotes may reveal target personas

External Frameworks

• Competitive intelligence frameworks
• Strategic analysis methodologies

Dean's Work

• Executive Insights Company Profile Template

Provenance

• Adapted from prompts/company-profile-executive-insights-research.md in the https://github.com/deanpeters/product-manager-prompts repo.

Skill type: Component Suggested filename: company-research.md Suggested placement: /skills/components/ Dependencies: References skills/positioning-statement/SKILL.md, skills/pestel-analysis/SKILL.md

Purpose

Guide product managers through diagnosing whether they're doing context stuffing (jamming volume without intent) or context engineering (shaping structure for attention). Use this to identify context boundaries, fix "Context Hoarding Disorder," and implement tactical practices like bounded domains, episodic retrieval, and the Research→Plan→Reset→Implement cycle.

Key Distinction: Context stuffing assumes volume = quality ("paste the entire PRD"). Context engineering treats AI attention as a scarce resource and allocates it deliberately.

This is not about prompt writing—it's about designing the information architecture that grounds AI in reality without overwhelming it with noise.

Input

Works best with: A description of the AI workflow, agent, or prompt setup that feels bloated, brittle, or hard to steer. Also useful: What you've already stuffed into context (docs, transcripts, schemas) and where outputs go wrong.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The advisor opens by asking what you're feeding the model today and what breaks.

Example invocation: Diagnose my setup: our support-triage agent gets the full 40-page policy manual per ticket and still misroutes edge cases.

Key Concepts

The Paradigm Shift: Parametric → Contextual Intelligence

The Fundamental Problem:

• LLMs have parametric knowledge (encoded during training) = static, outdated, non-attributable
• When asked about proprietary data, real-time info, or user preferences → forced to hallucinate or admit ignorance
• Context engineering bridges the gap between static training and dynamic reality

PM's Role Shift: From feature builder → architect of informational ecosystems that ground AI in reality

Context Stuffing vs. Context Engineering

Critical Metaphor: Context stuffing is like bringing your entire file cabinet to a meeting. Context engineering is bringing only the 3 documents relevant to today's decision.

The Anti-Pattern: Context Stuffing

Five Markers of Context Stuffing:

1. Reflexively expanding context windows — "Just add more tokens!"
2. Persisting everything "just in case" — No clear retention criteria
3. Chaining agents without boundaries — Agent A passes everything to Agent B to Agent C
4. Adding evaluations to mask inconsistency — "We'll just retry until it's right"
5. Normalized retries — "It works if you run it 3 times" becomes acceptable

Why It Fails:

• Reasoning Noise: Thousands of irrelevant files compete for attention, degrading multi-hop logic
• Context Rot: Dead ends, past errors, irrelevant data accumulate → goal drift
• Lost in the Middle: Models prioritize beginning (primacy) and end (recency), ignore middle
• Economic Waste: Every query becomes expensive without accuracy gains
• Quantitative Degradation: Accuracy drops below 20% when context exceeds ~32k tokens

The Hidden Costs:

• Escalating token consumption
• Diluted attention across irrelevant material
• Reduced output confidence
• Cascading retries that waste time and money

Real Context Engineering: Core Principles

Five Foundational Principles:

1. Context without shape becomes noise
2. Structure > Volume
3. Retrieve with intent, not completeness
4. Small working contexts (like short-term memory)
5. Context Compaction: Maximize density of relevant information per token

Quantitative Framework:

Efficiency = (Accuracy × Coherence) / (Tokens × Latency)

Key Finding: Using RAG with 25% of available tokens preserves 95% accuracy while significantly reducing latency and cost.

The 5 Diagnostic Questions (Detect Context Hoarding Disorder)

Ask these to identify context stuffing:

1. What specific decision does this support? — If you can't answer, you don't need it
2. Can retrieval replace persistence? — Just-in-time beats always-available
3. Who owns the context boundary? — If no one, it'll grow forever
4. What fails if we exclude this? — If nothing breaks, delete it
5. Are we fixing structure or avoiding it? — Stuffing context often masks bad information architecture

Memory Architecture: Two-Layer System

Short-Term (Conversational) Memory:

• Immediate interaction history for follow-up questions
• Challenge: Space management → older parts summarized or truncated
• Lifespan: Single session

Long-Term (Persistent) Memory:

• User preferences, key facts across sessions → deep personalization
• Implemented via vector database (semantic retrieval)
• Two types:
  • Declarative Memory: Facts ("I'm vegan")
  • Procedural Memory: Behavioral patterns ("I debug by checking logs first")
• Lifespan: Persistent across sessions

LLM-Powered ETL: Models generate their own memories by identifying signals, consolidating with existing data, updating database automatically.

The Research → Plan → Reset → Implement Cycle

The Context Rot Solution:

1. Research: Agent gathers data → large, chaotic context window (noise + dead ends)
2. Plan: Agent synthesizes into high-density SPEC.md or PLAN.md (Source of Truth)
3. Reset: Clear entire context window (prevents context rot)
4. Implement: Fresh session using only the high-density plan as context

Why This Works: Context rot is eliminated; agent starts clean with compressed, high-signal context.

Anti-Patterns (What This Is NOT)

• Not about choosing AI tools — Claude vs. ChatGPT doesn't matter; architecture matters
• Not about writing better prompts — This is systems design, not copywriting
• Not about adding more tokens — "Infinite context" narratives are marketing, not engineering reality
• Not about replacing human judgment — Context engineering amplifies judgment, doesn't eliminate it

When to Use This Skill

✅ Use this when:

• You're pasting entire PRDs/codebases into AI and getting vague responses
• AI outputs are inconsistent ("works sometimes, not others")
• You're burning tokens without seeing accuracy improvements
• You suspect you're "context stuffing" but don't know how to fix it
• You need to design context architecture for an AI product feature

❌ Don't use this when:

• You're just getting started with AI (start with basic prompts first)
• You're looking for tool recommendations (this is about architecture, not tooling)
• Your AI usage is working well (if it ain't broke, don't fix it)

Facilitation Source of Truth

Use workshop-facilitation as the default interaction protocol for this skill.

It defines:

• session heads-up + entry mode (Guided, Context dump, Best guess)
• one-question turns with plain-language prompts
• progress labels (for example, Context Qx/8 and Scoring Qx/5)
• interruption handling and pause/resume behavior
• numbered recommendations at decision points
• quick-select numbered response options for regular questions (include Other (specify) when useful)

This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.

Application

This interactive skill uses adaptive questioning to diagnose context stuffing, identify boundaries, and provide tactical implementation guidance.

Step 0: Gather Context

Agent asks:

Before we diagnose your context practices, let's gather information:

Current AI Usage:

• What AI tools/systems do you use? (ChatGPT, Claude, custom agents, etc.)
• What PM tasks do you use AI for? (PRD writing, user research synthesis, discovery, etc.)
• How do you provide context? (paste docs, reference files, use projects/memory)

Symptoms:

• Are AI outputs inconsistent? (works sometimes, not others)
• Are you retrying prompts multiple times to get good results?
• Are responses vague or hedged despite providing "all the context"?
• Are token costs escalating without accuracy improvements?

System Architecture (if applicable):

• Do you have custom AI agents or workflows?
• How is context shared between agents?
• Do you use RAG, vector databases, or memory systems?

You can describe briefly or paste examples.

Step 1: Diagnose Context Stuffing Symptoms

Agent asks:

Let's assess whether you're experiencing context stuffing. Which of these symptoms do you recognize?

Select all that apply:

1. "I paste entire documents into AI" — Full PRDs, complete user interview transcripts, entire codebases
2. "AI gives vague, hedged responses despite having 'all the context'" — Responses like "it depends," "consider these options," non-committal
3. "I have to retry prompts 3+ times to get usable output" — Inconsistency is normalized
4. "Token costs are escalating but accuracy isn't improving" — Spending more, getting same or worse results
5. "I keep adding more context hoping it'll help" — Reflexive expansion without strategy
6. "My agents pass everything to each other" — Agent A → Agent B → Agent C with full context chain
7. "I don't have clear criteria for what to include/exclude" — No context boundary definitions
8. "None of these—my AI usage is working well" — Skip to advanced optimization

User response: [Select symptoms]

Agent analyzes:

Based on your selections:

• 0-1 symptoms: Healthy context practices; proceed to optimization
• 2-3 symptoms: Early context stuffing; address before it scales
• 4+ symptoms: Active Context Hoarding Disorder; immediate intervention needed

Agent proceeds to diagnostic questions.

Step 2: Diagnostic Question 1 — What Specific Decision Does This Support?

Agent asks:

Let's start with the most critical question: For each piece of context you're including, what specific decision does it support?

Example Context Analysis:

Imagine you're asking AI to "help with discovery planning." You're providing:

• Entire PRD (20 pages)
• 50 user interview transcripts (full)
• Competitive analysis doc (15 pages)
• Team meeting notes from last 3 months

Question: What decision are you making right now?

Offer 3 scenarios:

1. "I'm deciding which user segment to interview first"

   • Context needed: User segments from PRD (2 paragraphs), prior interview themes (1 page synthesis), not full transcripts
   • Context NOT needed: Meeting notes, full competitive analysis, full PRD

2. "I'm deciding which discovery questions to ask in interviews"

   • Context needed: Research objectives (from PRD), past interview insights (synthesis), Jobs-to-be-Done framework
   • Context NOT needed: Full competitive analysis, full meeting notes

3. "I'm not sure what decision I'm making—I just want AI to 'understand my product'"

   • Problem: No specific decision = context stuffing trap
   • Fix: Define the decision first, then select context

Agent recommends:

Best Practice: Before adding context, complete this sentence:

"I need this context because I'm deciding [specific decision], and without [specific information], I can't make that decision."

If you can't complete that sentence, you don't need the context.

User response: [Describe their decision + context]

Agent validates: Does the context directly support the stated decision? If not, recommend trimming.

Step 3: Diagnostic Question 2 — Can Retrieval Replace Persistence?

Agent asks:

Second question: Is this information you always need, or something you can retrieve just-in-time?

The Distinction:

Always-Needed (Persist):

• Core product constraints (technical, regulatory, strategic)
• User preferences that apply to every interaction
• Critical definitions (operational glossary)
• Non-negotiable rules

Episodic (Retrieve on-demand):

• Project-specific details (this epic, this sprint)
• Historical data (past PRDs, old interview transcripts)
• Contextual facts (competitive analysis, market research)
• Temporary decisions

Key Insight: Just-in-time retrieval beats always-available. Don't persist what you can retrieve.

Offer 3 options:

1. "Most of my context is always-needed (core constraints, user prefs)"

   • Assessment: Good instinct; verify with Question 4 (what fails if excluded?)
   • Recommendation: Build constraints registry and operational glossary (persist these)

2. "Most of my context is episodic (project details, historical data)"

   • Assessment: Perfect candidate for RAG or retrieval
   • Recommendation: Implement semantic search; retrieve only relevant chunks for each query

3. "I'm not sure which is which—I persist everything to be safe"

   • Assessment: Classic Context Hoarding Disorder symptom
   • Fix: Apply Question 4 test to each piece of context

Agent recommends:

Rule of Thumb:

• Persist: Information referenced in 80%+ of interactions
• Retrieve: Information referenced in <20% of interactions
• Gray zone (20-80%): Depends on retrieval latency vs. context window cost

User response: [Categorize their context]

Agent provides: Specific recommendations on what to persist vs. retrieve.

Step 4: Diagnostic Question 3 — Who Owns the Context Boundary?

Agent asks:

Third question: Who is responsible for defining what belongs in vs. out of your AI's context?

The Ownership Problem:

If no one owns the context boundary, it will grow indefinitely. Every PM will add "just one more thing," and six months later, you're stuffing 100k tokens per query.

Offer 3 options:

1. "I own the boundary (solo PM or small team)"

   • Assessment: Good—you can make fast decisions
   • Recommendation: Document your boundary criteria (use Questions 1-5 as framework)

2. "My team shares ownership (collaborative boundary definition)"

   • Assessment: Can work if formalized
   • Recommendation: Create a "Context Manifest" doc: what's always included, what's retrieved, what's excluded (and why)

3. "No one owns it—it's ad-hoc / implicit"

   • Assessment: Critical risk; boundary will expand uncontrollably
   • Fix: Assign explicit ownership; schedule quarterly context audits

Agent recommends:

Best Practice: Create a Context Manifest

# Context Manifest: [Product/Feature Name]

## Always Persisted (Core Context)
- Product constraints (technical, regulatory)
- User preferences (role, permissions, preferences)
- Operational glossary (20 key terms)

## Retrieved On-Demand (Episodic Context)
- Historical PRDs (retrieve via semantic search)
- User interview transcripts (retrieve relevant quotes)
- Competitive analysis (retrieve when explicitly needed)

## Excluded (Out of Scope)
- Meeting notes older than 30 days (no longer relevant)
- Full codebase (use code search instead)
- Marketing materials (not decision-relevant)

## Boundary Owner: [Name]
## Last Reviewed: [Date]
## Next Review: [Date + 90 days]

User response: [Describe current ownership model]

Agent provides: Recommendation on formalizing ownership + template for Context Manifest.

Step 5: Diagnostic Question 4 — What Fails if We Exclude This?

Agent asks:

Fourth question: For each piece of context, what specific failure mode occurs if you exclude it?

This is the falsification test. If you can't identify a concrete failure, you don't need the context.

Offer 3 scenarios:

1. "If I exclude product constraints, AI will recommend infeasible solutions"

   • Failure Mode: Clear and concrete
   • Assessment: Valid reason to persist constraints

2. "If I exclude historical PRDs, AI won't understand our product evolution"

   • Failure Mode: Vague and hypothetical
   • Assessment: Historical context rarely needed for current decisions
   • Fix: Retrieve PRDs only when explicitly referencing past decisions

3. "If I exclude this, I'm not sure anything would break—I just include it to be thorough"

   • Failure Mode: None identified
   • Assessment: Context stuffing; delete immediately

Agent recommends:

The Falsification Protocol:

For each context element, complete this statement:

"If I exclude [context element], then [specific failure] will occur in [specific scenario]."

Examples:

• ✅ Good: "If I exclude GDPR constraints, AI will recommend features that violate EU privacy law."
• ❌ Bad: "If I exclude this PRD, AI might not fully understand the product." (Vague)

User response: [Apply falsification test to their context]

Agent provides: List of context elements to delete (no concrete failure identified).

Step 6: Diagnostic Question 5 — Are We Fixing Structure or Avoiding It?

Agent asks:

Fifth question: Is adding more context solving a problem, or masking a deeper structural issue?

The Root Cause Question:

Context stuffing often hides bad information architecture. Instead of fixing messy, ambiguous documents, teams add more documents hoping AI will "figure it out."

Offer 3 options:

1. "I'm adding context because our docs are poorly structured/ambiguous"

   • Assessment: You're masking a structural problem
   • Fix: Clean up the docs first (remove ambiguity, add constraints, define terms)
   • Example: Instead of pasting 5 conflicting PRDs, reconcile them into 1 Source of Truth

2. "I'm adding context because we don't have a shared operational glossary"

   • Assessment: You're compensating for missing foundations
   • Fix: Build the glossary (20-30 key terms); AI can reference it reliably
   • Example: Define "active user," "churn," "engagement" unambiguously

3. "I'm adding context because our constraints aren't documented"

   • Assessment: You're avoiding constraint engineering
   • Fix: Create constraints registry (technical, regulatory, strategic)
   • Example: Document "We won't build mobile apps" vs. explaining it in every prompt

Agent recommends:

The Structural Health Test:

If you're adding context to compensate for:

• Ambiguous documentation → Fix the docs, don't add more
• Undefined terms → Build operational glossary
• Undocumented constraints → Create constraints registry
• Conflicting information → Reconcile into Source of Truth

User response: [Identify structural issues]

Agent provides: Prioritized list of structural fixes before adding more context.

Step 7: Define Memory Architecture

Agent asks:

Based on your context needs, let's design a two-layer memory architecture.

Your Context Profile (from previous steps):

• Always-needed context: [Summary from Q2]
• Episodic context: [Summary from Q2]
• Boundary owner: [From Q3]
• Validated essentials: [From Q4]
• Structural fixes needed: [From Q5]

Recommended Architecture:

Short-Term (Conversational) Memory:

• What it stores: Immediate interaction history for follow-up questions
• Lifespan: Single session
• Management: Summarize or truncate older parts to avoid crowding
• Your specific needs: [Agent customizes based on user's workflow]

Long-Term (Persistent) Memory:

• What it stores: User preferences, core constraints, operational glossary
• Lifespan: Persistent across sessions
• Implementation: Vector database (semantic retrieval)
• Two types:
  • Declarative Memory: Facts (e.g., "We follow HIPAA regulations")
  • Procedural Memory: Behavioral patterns (e.g., "Always validate feasibility before usability")
• Your specific needs: [Agent customizes]

Retrieval Strategy (Episodic Context):

• What it retrieves: Historical PRDs, user interviews, competitive analysis
• Method: Semantic search triggered by query intent
• Optimization: Contextual Retrieval (Anthropic) — prepend explanatory context to each chunk before embedding
• Your specific needs: [Agent customizes]

Agent offers:

Would you like me to:

1. Generate a Context Architecture Blueprint for your specific use case?
2. Provide implementation guidance (tools, techniques, best practices)?
3. Design a retrieval strategy for your episodic context?

User response: [Selection]

Step 8: Implement Research → Plan → Reset → Implement Cycle

Agent asks:

Now let's implement the Research → Plan → Reset → Implement cycle to prevent context rot.

The Problem: As agents research, context windows grow chaotic—filled with dead ends, errors, and noise. This dilutes attention and causes goal drift.

The Solution: Compress research into a high-density plan, then clear the context window before implementing.

The Four-Phase Cycle:

Phase 1: Research (Chaotic Context Allowed)

• Agent gathers data from multiple sources
• Context window grows large and messy (this is expected)
• Dead ends, failed hypotheses, and noise accumulate
• Goal: Comprehensive information gathering

Phase 2: Plan (Synthesis)

• Agent synthesizes research into a high-density SPEC.md or PLAN.md
• This becomes the Source of Truth for implementation
• Key elements:
  • Decision made
  • Evidence supporting decision
  • Constraints applied
  • Next steps (sequenced)
• Format: Structured, concise, unambiguous

Phase 3: Reset (Clear Context Window)

• Critical step: Clear the entire context window
• Delete all research artifacts, dead ends, errors
• This prevents context rot from poisoning implementation

Phase 4: Implement (Fresh Session with Plan Only)

• Start a new session with only the high-density plan as context
• Agent has clean, focused attention on execution
• No noise from research phase

Agent offers 3 options:

1. "I want a template for the PLAN.md format"

   • Agent provides structured template for high-density plans

2. "I want to see an example of this cycle in action"

   • Agent walks through concrete PM use case (e.g., discovery planning)

3. "I'm ready to implement this in my workflow"

   • Agent provides step-by-step implementation guide

User response: [Selection]

Agent provides: Tailored guidance based on selection.

Step 9: Action Plan & Next Steps

Agent synthesizes:

Based on your context engineering assessment, here's your action plan:

Immediate Fixes (This Week):

1. [Delete context with no falsifiable failure mode from Q4]
2. [Apply Research→Plan→Reset→Implement to your next AI task]
3. [Document context boundary in Context Manifest]

Foundation Building (Next 2 Weeks):

1. [Build constraints registry with 20+ entries]
2. [Create operational glossary with 20-30 key terms]
3. [Implement two-layer memory architecture]

Long-Term Optimization (Next Month):

1. [Set up semantic retrieval for episodic context]
2. [Assign context boundary owner + quarterly audit schedule]
3. [Implement Contextual Retrieval (Anthropic) for RAG]

Success Metrics:

• Token usage down 50%+ (less context stuffing)
• Output consistency up (less retry/regeneration)
• Response quality up (sharper, less hedged answers)
• Context window stable (no unbounded growth)

Agent offers:

Would you like me to:

1. Generate specific implementation docs (Context Manifest, PLAN.md template, etc.)?
2. Provide advanced techniques (Contextual Retrieval, LLM-powered ETL)?
3. Review your current context setup (provide feedback on specific prompts/workflows)?

Examples

Example 1: Solo PM Context Stuffing → Engineering

Context:

• Solo PM at early-stage startup
• Using Claude Projects for PRD writing
• Pasting entire PRDs (20 pages) + all user interviews (50 transcripts) every time
• Getting vague, inconsistent responses

Assessment:

• Symptoms: Hedged responses, normalized retries (4+ symptoms)
• Q1 (Decision): "I just want AI to understand my product" (no specific decision)
• Q2 (Persist/Retrieve): Persisting everything (no retrieval strategy)
• Q3 (Ownership): No formal owner (solo PM, ad-hoc)
• Q4 (Failure): Can't identify concrete failures for most context
• Q5 (Structure): Avoiding constraint documentation

Diagnosis: Active Context Hoarding Disorder

Intervention:

1. Immediate: Delete all context that fails Q4 test → keeps 20% of original
2. Week 1: Build constraints registry (10 technical constraints, 5 strategic)
3. Week 2: Create operational glossary (25 terms)
4. Week 3: Implement Research→Plan→Reset→Implement for next PRD

Outcome: Token usage down 70%, output quality up significantly, responses crisp and actionable.

Example 2: Growth-Stage Team with Agent Chains

Context:

• Product team with 5 PMs
• Custom AI agents for discovery synthesis
• Agent A (research) → Agent B (synthesis) → Agent C (recommendations)
• Each agent passes full context to next → context window explodes to 100k+ tokens

Assessment:

• Symptoms: Escalating token costs, inconsistent outputs (3 symptoms)
• Q1 (Decision): Each agent has clear decision, but passes unnecessary context
• Q2 (Persist/Retrieve): Mixing persistent and episodic without strategy
• Q3 (Ownership): No explicit owner; each PM adds context
• Q4 (Failure): Agents pass "just in case" context with no falsifiable failure
• Q5 (Structure): Missing Context Manifest

Diagnosis: Agent orchestration without boundaries

Intervention:

1. Immediate: Define bounded context per agent (Agent A outputs only 2-page synthesis to Agent B, not full research)
2. Week 1: Assign context boundary owner (Lead PM)
3. Week 2: Create Context Manifest (what persists, what's retrieved, what's excluded)
4. Week 3: Implement Research→Plan→Reset→Implement between Agent B and Agent C

Outcome: Token usage down 60%, agent chain reliability up, costs reduced by 50%.

Example 3: Enterprise with RAG but No Context Engineering

Context:

• Large enterprise with vector database RAG system
• "Stuff the whole knowledge base" approach (10,000+ documents)
• Retrieval returns 50+ chunks per query → floods context window
• Accuracy declining as knowledge base grows

Assessment:

• Symptoms: Vague responses despite "complete knowledge," normalized retries (2 symptoms)
• Q1 (Decision): Decisions clear, but retrieval has no intent (returns everything)
• Q2 (Persist/Retrieve): Good instinct to retrieve, but no filtering
• Q3 (Ownership): Engineering owns RAG, Product doesn't own context boundaries
• Q4 (Failure): Can't identify why 50 chunks needed vs. 5
• Q5 (Structure): Knowledge base has no structure (flat documents, no metadata)

Diagnosis: Retrieval without intent (RAG as context stuffing)

Intervention:

1. Immediate: Limit retrieval to top 5 chunks per query (down from 50)
2. Week 1: Implement Contextual Retrieval (Anthropic) — prepend explanatory context to each chunk during indexing
3. Week 2: Add metadata to documents (category, recency, authority)
4. Week 3: Product team defines retrieval intent per query type (discovery = customer insights, feasibility = technical constraints)

Outcome: Accuracy up 35% (from Anthropic benchmark), latency down 60%, token usage down 80%.

Common Pitfalls

1. "Infinite Context" Marketing vs. Engineering Reality

Failure Mode: Believing "1 million token context windows" means you should use all of them.

Consequence: Reasoning Noise degrades performance; accuracy drops below 20% past ~32k tokens.

Fix: Context windows are not free. Treat tokens as scarce; optimize for density, not volume.

2. Retrying Instead of Restructuring

Failure Mode: "It works if I run it 3 times" → normalizing retries instead of fixing structure.

Consequence: Wastes time and money; masks deeper context rot issues.

Fix: If retries are common, your context structure is broken. Apply Q5 (fix structure, don't add volume).

3. No Context Boundary Owner

Failure Mode: Ad-hoc, implicit context decisions → unbounded growth.

Consequence: Six months later, every query stuffs 100k tokens per interaction.

Fix: Assign explicit ownership; create Context Manifest; schedule quarterly audits.

4. Mixing Always-Needed with Episodic

Failure Mode: Persisting historical data that should be retrieved on-demand.

Consequence: Context window crowded with irrelevant information; attention diluted.

Fix: Apply Q2 test: persist only what's needed in 80%+ of interactions; retrieve the rest.

5. Skipping the Reset Phase

Failure Mode: Never clearing context window during Research→Plan→Implement cycle.

Consequence: Context rot accumulates; goal drift; dead ends poison implementation.

Fix: Mandatory Reset phase after Plan; start implementation with only high-density plan as context.

References

Related Skills

• ai-shaped-readiness-advisor (Interactive) — Context Design is Competency #1 of AI-shaped work
• problem-statement (Component) — Evidence-based framing requires context engineering
• epic-hypothesis (Component) — Testable hypotheses depend on clear constraints (part of context)
• pol-probe-advisor (Interactive) — Validation experiments benefit from context engineering (define what AI needs to know)

External Frameworks

• Dean Peters — Context Stuffing Is Not Context Engineering (Dean Peters' Substack, 2026)
• Teresa Torres — Continuous Discovery Habits (Context Engineering as one of 5 new AI PM disciplines)
• Marty Cagan — Empowered (Feasibility risk in AI era includes understanding "physics of AI")
• Anthropic — Contextual Retrieval whitepaper (35% failure rate reduction)
• Google — Context engineering whitepaper on LLM-powered memory systems

Technical References

• RAG (Retrieval-Augmented Generation) — Standard technique for episodic context retrieval
• Vector Databases — Semantic search for long-term memory (Pinecone, Weaviate, Chroma)
• Contextual Retrieval (Anthropic) — Prepend explanatory context to chunks before embedding
• LLM-as-Judge — Automated evaluation of context quality

Purpose

Create a comprehensive customer journey map that visualizes how customers interact with your brand across all stages—from awareness to loyalty—documenting their actions, touchpoints, emotions, KPIs, business goals, and teams involved at each stage. Use this to identify pain points, align cross-functional teams, and systematically improve the customer experience to achieve business objectives.

This is not a user flow diagram—it's a strategic artifact that combines customer empathy with business metrics to drive actionable improvements.

Input

Works best with: The customer (persona or segment) and the scenario or goal the journey covers. Also useful: Known stages, touchpoints, pain points, supporting research, and the business objective behind the map.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The skill asks for persona and scenario first — a journey map without a specific actor and goal turns into a generic funnel diagram.

Example invocation: Map the journey for a mid-market ops manager going from free trial to paid rollout across their team.

Key Concepts

The Customer Journey Mapping Framework

Adapted from NNGroup's framework and Carnegie Mellon's PM curriculum, a customer journey map documents:

Horizontal structure (stages):

• Awareness: Customer first learns about your brand
• Consideration: Customer evaluates your offering
• Decision: Customer makes a purchase
• Service: Customer uses the product/service post-purchase
• Loyalty: Customer becomes a repeat buyer and advocate

Vertical structure (for each stage):

• Customer Actions: What customers do
• Touchpoints: Where/how they interact with your brand
• Customer Experience: Emotions and thoughts
• KPIs: Metrics to measure success
• Business Goals: What you're trying to achieve
• Teams Involved: Who owns this stage

Why This Works

• Empathy-driven: Centers on customer emotions, not just actions
• Cross-functional alignment: Shows which teams affect which stages
• Metric-focused: Ties customer experience to measurable outcomes
• Gap identification: Makes pain points and opportunities visible
• Actionable: Clear KPIs and goals enable prioritization

Anti-Patterns (What This Is NOT)

• Not a user story map: Journey maps are broader (all touchpoints, not just product use)
• Not a service blueprint: Less detailed on internal processes, more focused on customer experience
• Not static: Journey maps evolve as customer behavior changes

When to Use This

• Understanding customer experience across all touchpoints (not just product)
• Aligning cross-functional teams (marketing, sales, product, support)
• Identifying pain points and prioritizing improvements
• Onboarding new team members to customer perspective
• Auditing the end-to-end customer experience

When NOT to Use This

• For deep product-specific workflows (use story mapping instead)
• Before defining personas (need to know who you're mapping)
• As a one-time exercise (journey maps require ongoing updates)

Application

Use template.md for the full fill-in structure.

Step 1: Prepare Prerequisites

Before mapping, ensure you have:

1. Key stakeholders: Marketing, sales, product, customer service representatives
2. Buyer personas: Detailed personas with demographics, psychographics, goals, challenges (reference skills/proto-persona/SKILL.md)
3. Defined stages: Main stages of your buying process (typically: Awareness, Consideration, Decision, Service, Loyalty)
4. Touchpoint inventory: All places customers interact with your brand (website, social, email, store, support, etc.)

If missing: Run discovery interviews, persona definition work, or touchpoint audits first.

Step 2: Set Clear Objectives

Define what you want to achieve:

## Objectives
- [Goal 1: e.g., "Identify top 3 pain points causing drop-off between Awareness and Consideration"]
- [Goal 2: e.g., "Align marketing and sales on customer motivations at each stage"]
- [Goal 3: e.g., "Understand emotional journey to inform messaging strategy"]

Quality checks:

• Specific: Not "understand customers" but "identify drop-off causes in Consideration stage"
• Actionable: Results should inform decisions, not just document observations

Step 3: Choose a Buyer Persona

Select one persona to focus on (create separate maps for each persona):

## Persona
- [Persona name and brief description]
- [Example: "Manager Mike: 35-42, Director of Product at mid-sized B2B SaaS, struggles with data-driven prioritization, values time savings over feature depth"]

Why one persona per map: Different personas have different journeys. Mixing them creates confusion.

Step 4: Map Each Stage

For each stage (Awareness, Consideration, Decision, Service, Loyalty), document:

Customer Actions

What customers do at this stage:

### Stage: [Stage Name, e.g., Awareness]

**Customer Actions:**
- [Action 1: e.g., "See LinkedIn ad about product management tools"]
- [Action 2: e.g., "Hear about tool from PM peer at conference"]
- [Action 3: e.g., "Google 'best product roadmap software'"]

Quality checks:

• Observable: You can see or measure this action
• Specific: Not "research products" but "Google 'best roadmap software' and read comparison articles"

Touchpoints

Where/how customers interact with your brand:

**Touchpoints:**
- [Touchpoint 1: e.g., "LinkedIn Ads"]
- [Touchpoint 2: e.g., "Word-of-mouth at PM conferences"]
- [Touchpoint 3: e.g., "Google organic search results"]
- [Touchpoint 4: e.g., "Review sites (G2, Capterra)"]

Quality checks:

• Comprehensive: Include both digital and physical touchpoints
• Specific: Not "social media" but "LinkedIn Ads," "Twitter mentions," etc.

Customer Experience

Emotions and thoughts customers have:

**Customer Experience:**
- [Emotion 1: e.g., "Curious but skeptical—'Is this actually better than spreadsheets?'"]
- [Emotion 2: e.g., "Overwhelmed by options—'Too many tools, how do I choose?'"]
- [Emotion 3: e.g., "Hopeful but cautious—'Could this save me time?'"]

Quality checks:

• Authentic: Use customer quotes from research when possible
• Emotional: Capture feelings, not just thoughts
• Specific: Not "interested" but "curious but skeptical—worried about setup time"

KPIs

Key performance indicators for this stage:

**KPIs:**
- [KPI 1: e.g., "Brand awareness (measured via surveys)"]
- [KPI 2: e.g., "LinkedIn ad impressions: 100k/month"]
- [KPI 3: e.g., "Organic search traffic: 5k visitors/month"]
- [KPI 4: e.g., "G2 review views: 2k/month"]

Quality checks:

• Measurable: Can you track this?
• Stage-appropriate: Awareness KPIs differ from Decision KPIs

Business Goals

What you're trying to achieve at this stage:

**Business Goals:**
- [Goal 1: e.g., "Increase brand awareness among PMs at B2B SaaS companies"]
- [Goal 2: e.g., "Generate 500 qualified leads/month"]
- [Goal 3: e.g., "Position as top 3 roadmap tool in G2 rankings"]

Quality checks:

• Outcome-focused: Not "run ads" but "increase brand awareness"
• Aligned with stage: Don't expect conversions at Awareness stage

Teams Involved

Who owns this stage:

**Teams Involved:**
- [Team 1: e.g., "Marketing (ad campaigns, SEO)"]
- [Team 2: e.g., "Content (blog posts, comparison guides)"]
- [Team 3: e.g., "Customer Success (case studies, testimonials)"]

Quality checks:

• Cross-functional: Multiple teams usually touch each stage
• Specific roles: Not just "marketing" but "marketing (ad campaigns, SEO)"

Step 5: Visualize the Map

Create a table or visual diagram:

Step 6: Analyze and Prioritize

Review the map and ask:

1. Where are the biggest pain points? (Look for negative emotions + high drop-off rates)
2. Which stages have the weakest KPIs? (Prioritize low-performing stages)
3. Are teams aligned? (Do teams understand their role in each stage?)
4. What opportunities exist? (Where can small improvements create big impact?)

Prioritization criteria:

• Impact: How much would fixing this improve the customer experience?
• Feasibility: How easy is this to fix?
• Alignment: Does this support business goals?

Step 7: Test and Refine

• Update regularly: Customer behavior changes—revisit the map quarterly
• Validate with data: Use analytics, surveys, and customer interviews to confirm assumptions
• Track improvements: After making changes, measure impact on KPIs

Examples

See examples/sample.md for a full customer journey map example. See examples/meta-product-manager-skills.md for a meta dogfooding example mapping this repository's own customer journey.

Mini example excerpt:

| **Stage** | **Awareness** | **Consideration** | **Decision** |
| **Customer Actions** | Sees LinkedIn ad | Compares on G2 | Starts free trial |
| **Customer Experience** | Curious but skeptical | Overwhelmed | Anxious about setup |

Common Pitfalls

Pitfall 1: Generic Emotions

Symptom: "Customer feels happy" or "Customer is satisfied"

Consequence: No insight into why they feel that way or what to improve.

Fix: Be specific: "Relieved that setup took 30 minutes, not 3 hours as feared."

Pitfall 2: Missing Touchpoints

Symptom: Only documenting digital touchpoints (website, app)

Consequence: Miss offline interactions (conferences, word-of-mouth, support calls).

Fix: Include all touchpoints: physical, digital, human, and automated.

Pitfall 3: Internal Perspective

Symptom: Mapping what you want customers to do, not what they actually do

Consequence: Journey map reflects wishful thinking, not reality.

Fix: Validate with customer research, analytics, and support tickets.

Pitfall 4: No KPIs or Goals

Symptom: Journey map has actions and emotions but no metrics or business objectives

Consequence: No way to measure success or prioritize improvements.

Fix: Add KPIs and business goals for each stage. Make them measurable.

Pitfall 5: One-and-Done Exercise

Symptom: Journey map created once, never updated

Consequence: Map becomes outdated as customer behavior evolves.

Fix: Review quarterly. Update based on new data, product changes, or market shifts.

References

Related Skills

• skills/proto-persona/SKILL.md — Defines the persona for the journey map
• skills/jobs-to-be-done/SKILL.md — Informs customer actions and goals
• skills/problem-statement/SKILL.md — Identifies pain points at each stage
• skills/user-story-mapping/SKILL.md — Complementary (story mapping focuses on product usage, journey mapping covers all touchpoints)

External Frameworks

• NNGroup, Customer Journey Mapping (2016) — Foundational framework
• Carnegie Mellon University, Product Management Curriculum — Academic approach
• Chris Risdon & Patrick Quattlebaum, Orchestrating Experiences (2018) — Journey mapping for service design

Dean's Work

• Customer Journey Mapping Prompt Template (adapted from NNGroup and CMU frameworks)

Provenance

• Adapted from prompts/customer-journey-mapping-prompt-template.md in the https://github.com/deanpeters/product-manager-prompts repo.

Skill type: Component Suggested filename: customer-journey-map.md Suggested placement: /skills/components/ Dependencies: References skills/proto-persona/SKILL.md, skills/jobs-to-be-done/SKILL.md, skills/problem-statement/SKILL.md

Purpose

Guide product managers through creating a customer journey map by asking adaptive questions about the actor (persona), scenario/goal, journey phases, actions/emotions, and opportunities for improvement. Use this to visualize the end-to-end customer experience, identify pain points, and create a shared mental model across teams—avoiding surface-level feature lists and ensuring discovery work focuses on real customer problems, not assumed solutions.

This is not a feature roadmap—it's a discovery and alignment tool that uncovers where the experience breaks down and where improvements will have the greatest impact.

Input

Works best with: The persona (actor) and the scenario or goal to map. Also useful: Journey phases you already know, research artifacts, and where you suspect the experience breaks.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The workshop opens with the actor question (Q1) and proceeds one question at a time per the facilitation protocol.

Example invocation: Run a journey mapping workshop: persona is a first-time landlord, scenario is listing and leasing their first property through our app.

Key Concepts

What is a Customer Journey Map?

A journey map (NNGroup) visualizes "the process that a person goes through in order to accomplish a goal." It compiles user actions into a timeline, enriched with thoughts and emotions to create a narrative, then condenses and polishes into a visual artifact.

Five Key Components (NNGroup Framework)

1. Actor — A specific persona or user whose perspective anchors the map
2. Scenario + Expectations — The situational context and associated goals
3. Journey Phases — High-level stages organizing the experience (e.g., discover, try, buy, use, seek support)
4. Actions, Mindsets, and Emotions — User behaviors, thoughts, and emotional responses throughout phases
5. Opportunities — Insights identifying where experience can improve

Journey Map Structure

Actor: [Persona Name]
Scenario: [Goal/Context]

Phase 1: Discover → Phase 2: Try → Phase 3: Buy → Phase 4: Use → Phase 5: Support
   ↓                  ↓                ↓               ↓               ↓
Actions:           Actions:         Actions:        Actions:        Actions:
Thoughts:          Thoughts:        Thoughts:       Thoughts:       Thoughts:
Emotions: 😊😐😞    Emotions:        Emotions:       Emotions:       Emotions:
   ↓                  ↓                ↓               ↓               ↓
Opportunities:     Opportunities:   Opportunities:  Opportunities:  Opportunities:

Why This Works

• Forces conversation: Teams align on shared understanding of customer experience
• Reveals pain points: Emotions + actions highlight where experience breaks down
• Prioritizes improvements: Opportunities ranked by impact guide roadmap decisions
• Human-centered: Focuses on customer perspective, not internal processes

Anti-Patterns (What This Is NOT)

• Not a service blueprint: Journey maps focus on customer perspective; service blueprints map internal operations
• Not a user story map: Journey maps support discovery; user story maps facilitate implementation planning
• Not an experience map: Journey maps target specific users and products; experience maps explore broader human behaviors

When to Use This

• Starting customer discovery (understanding current experience)
• Identifying pain points for retention/engagement initiatives
• Aligning cross-functional teams on customer perspective
• Prioritizing which problems to solve first

When NOT to Use This

• When you already understand the customer journey deeply
• For technical refactoring (no customer-facing journey)
• As a substitute for user research (maps require research input)

Facilitation Source of Truth

Use workshop-facilitation as the default interaction protocol for this skill.

It defines:

• session heads-up + entry mode (Guided, Context dump, Best guess)
• one-question turns with plain-language prompts
• progress labels (for example, Context Qx/8 and Scoring Qx/5)
• interruption handling and pause/resume behavior
• numbered recommendations at decision points
• quick-select numbered response options for regular questions (include Other (specify) when useful)

This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.

Application

This interactive skill asks up to 5 adaptive questions, offering 3-4 enumerated options at each step.

Interaction pattern: Pair with skills/workshop-facilitation/SKILL.md when you want a one-step-at-a-time flow with numbered recommendations at decision points and quick-select options for regular questions. If the user asks for a single-shot output, skip the multi-turn facilitation.

Step 0: Gather Context (Before Questions)

Agent suggests:

Before we create your journey map, let's gather context:

Customer Research:

• User interviews, discovery notes, support tickets
• Churn reasons, exit surveys, NPS feedback
• Analytics data (drop-off points, feature usage)
• Personas or proto-personas

Product Context:

• Website copy, product descriptions, positioning
• Competitor journey maps or reviews (G2, Capterra)
• Existing journey documentation (if any)

You can paste this content directly, or describe the customer experience briefly.

Question 1: Identify Actor (Persona)

Agent asks: "Who is the actor for this journey map? (Which persona or user segment?)"

Offer 4 enumerated options:

1. Primary persona — "Your main target customer (e.g., 'small business owner')" (Most common starting point)
2. Secondary persona — "A secondary user segment with different needs (e.g., 'enterprise admin' vs. 'end user')" (Use if primary persona already mapped)
3. High-churn persona — "User segment with highest churn rate (e.g., 'trial users who don't convert')" (Good for retention initiatives)
4. Newly discovered persona — "Emerging user segment from recent research (e.g., 'remote teams' post-COVID)" (Good for market expansion)

Or describe your specific persona.

Adaptation: Use personas from context (proto-personas, JTBD research, etc.)

User response: [Selection or custom]

Question 2: Define Scenario + Goal

Agent asks: "What's the scenario and goal for this journey? (What is the actor trying to accomplish?)"

Offer 4 enumerated options:

1. First-time use — "New user onboarding, from discovery to activation" (Common for SaaS, apps)
2. Core workflow — "Recurring task the user does regularly (e.g., 'create invoice,' 'run report')" (Common for established products)
3. Problem resolution — "User encounters issue and seeks help (e.g., 'forgot password,' 'billing question')" (Good for support/retention)
4. Upgrade/expansion — "Free user considering paid plan, or existing customer expanding usage" (Good for growth initiatives)

Or describe your specific scenario.

User response: [Selection or custom]

Agent extracts:

• Actor: [Persona from Q1]
• Scenario: [Context from Q2]
• Goal: [What actor is trying to accomplish]

Question 3: Identify Journey Phases

Agent says: "Let's break the journey into high-level phases (typically 4-6 phases from start to end)."

Agent generates 4-6 journey phases based on scenario (Q2).

Example (if Scenario = "First-time use"):

Journey Phases (left to right):

1. Discover — User learns about product
2. Evaluate — User researches, compares alternatives
3. Try — User signs up, starts onboarding
4. Activate — User reaches "aha moment," experiences value
5. Use — User integrates product into workflow
6. Expand — User considers upgrading or inviting team

Agent asks: "Do these phases capture the full journey? Should we add, remove, or rename phases?"

User response: [Approve or modify]

Question 4: Map Actions, Thoughts, Emotions per Phase

Agent says: "Now let's map what the actor does, thinks, and feels in each phase."

Agent generates 3-5 actions, thoughts, and emotions per phase based on context (Step 0) and scenario (Q2).

Example (for Phase 3: "Try — User signs up, starts onboarding"):

Phase 3: Try (Onboarding)

Actions:
- Signs up with email
- Receives welcome email
- Logs in for the first time
- Sees empty dashboard
- Searches for "getting started" guide

Thoughts:
- "This looks promising, but I'm not sure where to start"
- "Do I need to watch a tutorial video?"
- "What's the first step?"

Emotions:
- Curious but uncertain 🤔
- Slightly frustrated (no clear next step) 😕
- Hopeful it will get easier 🙂

Pain Points:
- No onboarding checklist or guided tour
- Empty state doesn't suggest next action
- Too many options in navigation (overwhelming)

Agent repeats for all journey phases, showing full map.

Agent asks: "Does this capture the customer experience accurately? Should we adjust actions, thoughts, or emotions?"

User response: [Approve or modify]

Question 5: Identify Opportunities (Pain Points to Address)

Agent says: "Based on the journey map, let's identify opportunities for improvement—ranked by impact."

Agent generates 5-7 opportunities (pain points with highest emotional intensity or drop-off rates).

Example:

# Opportunities (Ranked by Impact)

## 1. Onboarding lacks guided first steps (Phase 3: Try)
**Pain Point:** Users see empty dashboard, don't know what to do first
**Evidence:** 60% of signups don't complete first action within 24 hours
**Opportunity:** Add interactive onboarding checklist ("Create your first project," "Invite a teammate")
**Impact:** HIGH — Directly affects activation rate

---

## 2. Pricing page is confusing (Phase 2: Evaluate)
**Pain Point:** Users don't understand which plan fits their needs
**Evidence:** High bounce rate on pricing page (70% leave without signing up)
**Opportunity:** Add plan comparison tool or "Which plan is right for me?" quiz
**Impact:** HIGH — Directly affects trial conversion

---

## 3. Support is hard to find (Phase 5: Use)
**Pain Point:** Users encounter issues, struggle to find help
**Evidence:** Support tickets often say "I couldn't find an answer in docs"
**Opportunity:** Add in-app help widget, contextual tooltips
**Impact:** MEDIUM — Affects retention, but fewer users hit this phase

---

## 4. Email confirmations lack context (Phase 1: Discover)
**Pain Point:** Marketing emails don't explain value clearly
**Evidence:** Low click-through rate on email campaigns (5% vs. industry avg 15%)
**Opportunity:** Rewrite emails with customer language, clear CTAs
**Impact:** MEDIUM — Affects top-of-funnel awareness

---

## 5. Upgrade prompts feel pushy (Phase 6: Expand)
**Pain Point:** Users perceive upgrade prompts as sales-y, not helpful
**Evidence:** Negative sentiment in NPS comments ("too many upgrade popups")
**Opportunity:** Show upgrade value contextually (when user hits free plan limit)
**Impact:** LOW — Affects smaller user subset

Agent asks: "Do these opportunities align with your priorities? Which should we focus on first?"

User response: [Selection or custom]

Output: Customer Journey Map + Opportunity List

After completing the flow, the agent outputs:

# Customer Journey Map: [Scenario from Q2]

**Actor:** [Persona from Q1]
**Scenario:** [Context from Q2]
**Goal:** [What actor is trying to accomplish]
**Date:** [Today's date]

---

## Journey Phases

[Phase 1] → [Phase 2] → [Phase 3] → [Phase 4] → [Phase 5] → [Phase 6]

---

## Full Journey Map

### Phase 1: [Name]

**Actions:**
- [Action 1]
- [Action 2]
- [Action 3]

**Thoughts:**
- "[Quote 1]"
- "[Quote 2]"

**Emotions:**
- [Emotion 1] 😊
- [Emotion 2] 😐

**Pain Points:**
- [Pain point 1]
- [Pain point 2]

---

### Phase 2: [Name]

[...repeat structure for all phases...]

---

## Opportunities (Prioritized)

### Opportunity 1: [Name] (HIGH IMPACT)
**Phase:** [Journey phase]
**Pain Point:** [Description]
**Evidence:** [Data/research]
**Proposed Solution:** [How to address]
**Impact:** HIGH — [Rationale]

---

### Opportunity 2: [Name] (HIGH IMPACT)
**Phase:** [Journey phase]
**Pain Point:** [Description]
**Evidence:** [Data/research]
**Proposed Solution:** [How to address]
**Impact:** HIGH — [Rationale]

---

[...continue for all opportunities...]

---

## Next Steps

1. **Validate opportunities:** Use `discovery-interview-prep.md` to test hypotheses with customers
2. **Prioritize fixes:** Use `prioritization-advisor.md` to choose which opportunities to tackle first
3. **Create problem statements:** Use `problem-statement.md` to frame top opportunities
4. **Build experiments:** Use `opportunity-solution-tree.md` to design solutions and POCs

---

**Ready to start addressing opportunities? Let me know if you'd like to refine the map or dive into a specific pain point.**

Examples

Example 1: Good Journey Map (SaaS Onboarding)

Q1 Response: "Primary persona — Small business owner"

Q2 Response: "First-time use — New user onboarding, from discovery to activation"

Q3 - Phases Generated:

Discover → Evaluate → Try → Activate → Use → Expand

Q4 - Phase 3 (Try) Mapped:

Actions:
- Signs up via Google SSO
- Receives welcome email
- Logs in, sees empty dashboard
- Clicks "Help" button, watches 5-min tutorial
- Attempts to create first project, gets stuck on form

Thoughts:
- "This looks easy enough"
- "Wait, what's a 'workspace' vs. 'project'?"
- "Do I need to fill out all these fields?"

Emotions:
- Excited initially 😊
- Confused by terminology 😕
- Frustrated by unclear form 😞

Pain Points:
- No guided onboarding checklist
- Terminology not explained (workspace vs. project)
- Form has too many required fields upfront

Q5 - Opportunities Identified:

1. Add onboarding checklist (HIGH — affects activation)
2. Simplify terminology (MEDIUM — affects understanding)
3. Reduce required form fields (MEDIUM — affects completion rate)

Why this works:

• Emotions + actions reveal pain points clearly
• Opportunities tied to specific phases
• Evidence from research (drop-off data, support tickets)

Example 2: Bad Journey Map (Too Generic)

Phase: "Use Product"

Actions:

• Uses product
• Does tasks

Thoughts:

• "This is good"

Emotions:

• Happy 😊

Why this fails:

• No specificity (what tasks? which features?)
• No pain points identified (everything is "good")
• Can't extract actionable opportunities

Fix:

• Get specific: "User creates invoice → sends to client → tracks payment status"
• Include real customer quotes: "I wish I could bulk-send invoices"
• Show emotional highs AND lows (not just happy)

Common Pitfalls

Pitfall 1: Mapping Internal Process, Not Customer Experience

Symptom: Journey phases = "Lead generated → Qualified → Demo scheduled → Deal closed"

Consequence: Focuses on sales process, not customer perspective

Fix: Map from customer POV: "Discovers problem → Researches solutions → Tries product → Adopts"

Pitfall 2: No Emotions or Pain Points

Symptom: Journey map lists actions only, no thoughts/emotions

Consequence: Misses the point—can't identify where experience breaks down

Fix: Add customer quotes, emotional states (frustrated, delighted, confused)

Pitfall 3: Too Many Personas in One Map

Symptom: Trying to map "all users" in a single journey

Consequence: Loses focus, becomes generic

Fix: One map per persona. If multiple personas, create separate maps.

Pitfall 4: Opportunities Aren't Prioritized

Symptom: List 20 opportunities with no ranking

Consequence: Team paralyzed, doesn't know where to start

Fix: Rank by impact (HIGH/MEDIUM/LOW) based on evidence and emotional intensity

Pitfall 5: Map Created in Isolation

Symptom: PM creates journey map alone, doesn't involve team

Consequence: No shared mental model, map doesn't drive decisions

Fix: Facilitate workshop with cross-functional team (PM, design, engineering, support)

References

Related Skills

• customer-journey-map.md — Component skill with journey map template
• proto-persona.md — Defines actor for journey mapping
• problem-statement.md — Converts opportunities into problem statements
• discovery-interview-prep.md — Gathers research input for mapping
• opportunity-solution-tree.md — Designs solutions for journey opportunities

External Frameworks

• Nielsen Norman Group, "Journey Mapping 101" (2016) — Definitive guide to journey mapping
• Adaptive Path, "Guide to Experience Mapping" (2013) — Experience vs. journey maps

Dean's Work

• [If Dean has journey mapping resources, link here]

Skill type: Interactive Suggested filename: customer-journey-mapping-workshop.md Suggested placement: /skills/interactive/ Dependencies: Uses customer-journey-map.md, proto-persona.md, problem-statement.md, jobs-to-be-done.md

Purpose

Guide product managers through identifying the most important things to measure, test, or track across 10 risk dimensions — 4 internal and 6 external — then triage each into act on immediately or start tracking. The output is a prioritized risk register that tells you what to do first, not just what could go wrong.

This is not a generic risk list. It's a guided scan tied to your specific product, customer, and market context. The question it answers: is the orange worth the squeeze?

Input

Works best with: The product or AI idea you want to stress-test. Also useful: Its stage (concept, prototype, in-market), what you've already validated, and any dimensions you already know are risky.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The advisor opens by asking for the idea in a sentence or two, then scans the 10 risk dimensions against it.

Example invocation: De-risk this: an AI agent that auto-drafts insurance claim responses for adjusters — prototype stage, no compliance review yet.

Key Concepts

Two Lenses, One Question

Every product idea carries two kinds of risk. Most PMs check one and ignore the other.

Internal risks ask whether the product itself will work:

• Product Outcome Metrics (Desirability + Usability) = "What's in it for the customer/user?"
• Business Outcome Metrics (Feasibility + Viability) = "What's in it for our business/organization?"

External risks ask whether the world will let it work:

• PESTEL forces (Political, Economic, Social, Technological, Environmental, Legal) = "What could blindside us?"

The single triage question that applies to both sides: Do I act on this now, or start tracking it?

The Internal Risk Framework (DUFV)

Adapted from Marty Cagan's four product risks, split into two measurement categories:

Product Outcome Metrics answer: What's in it for the customer/user? They measure whether the product is solving meaningful problems and delivering value to its intended users.

Business Outcome Metrics answer: What's in it for our business/organization? They measure how effectively and reliably the product supports a seamless customer experience while remaining feasible to deliver.

The connection to positioning: Product Outcome Metrics validate your Problem Positioning Statement (the customer promise). Business Outcome Metrics validate your Problem Framing Statement (the business case).

The External Risk Framework (PESTEL)

Six categories of forces outside your organization that can reshape, delay, or kill your product plans:

Key insight from the classroom: Market conditions don't just change — they potentially change us, our business, and our products. PESTEL helps PMs categorize and respond to that change, not just observe it.

The Act/Watch Triage

Every surfaced risk gets one of two labels:

• Act on immediately: This risk is present, material, and requires a response now — run a test, change a plan, mitigate a gap.
• Start tracking: This risk is real but not yet urgent — monitor it, set a trigger threshold, revisit on a cadence.

The triage applies equally to internal and external risks. A desirability gap you can't answer is an "act now" just as much as a regulatory deadline. A demographic trend you see coming is a "watch" just like a feasibility concern you've already partially mitigated.

Anti-pattern: Labeling everything "act now." If every risk is urgent, none of them are prioritized. Force yourself to split roughly 40/60 act/watch.

When to Use This Skill

Use this when:

• You have a product or AI idea and need to decide whether to commit
• You're about to put something on the roadmap and want a structured risk check
• You've been focused on internal risks and haven't scanned external forces (or vice versa)
• Leadership asks "what could go wrong?" and you want a better answer than "nothing, we're good"

Don't use this when:

• You don't have a specific product idea yet (use problem-framing-canvas first)
• You need a deep PESTEL analysis with data and citations (use pestel-analysis component skill)
• You've already validated the idea and need to pick a validation method (use pol-probe-advisor)
• The idea is already in production and you're optimizing, not de-risking

Facilitation Source of Truth

Use workshop-facilitation as the default interaction protocol for this skill.

It defines:

• session heads-up + entry mode (Guided, Context dump, Best guess)
• one-question turns with plain-language prompts
• progress labels (for example, Context Qx/4 and Risk Scan Qx/6)
• interruption handling and pause/resume behavior
• numbered recommendations at decision points
• quick-select numbered response options for regular questions (include Other (specify) when useful)

This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.

Application

This interactive skill uses adaptive questioning to scan your product idea across 10 risk dimensions (4 internal + 6 external), triage each surfaced risk into act/watch, and produce a prioritized risk register.

Session Start: Heads-Up + Entry Mode (Mandatory)

Agent opening prompt:

"Quick heads-up before we start: this usually takes about 15-20 minutes. We'll cover 4 context questions, then scan 10 risk dimensions — 4 internal (will customers want it, can they use it, can we build it, does the business work) and 6 external (political, economic, social, technological, environmental, legal). For each risk that surfaces, we'll triage it: act now or track.

How do you want to do this?

1. Guided mode: I'll walk through each dimension one at a time.
2. Context dump: Paste what you know about the idea and I'll skip what's covered.
3. Best guess mode: Give me the idea in a sentence and I'll make reasonable assumptions, labeled."

Step 0: Gather Context

Agent asks the following questions, one per turn:

Context Q1/4: "What's the product or AI idea you're evaluating? Describe it in 1-2 sentences — what it does and who it's for."

Context Q2/4: "What stage is this idea at?"

1. Concept — we're still exploring whether to pursue it
2. Prototype — we have something testable but uncommitted
3. Building — we've committed and are developing it
4. Scaling — it's live and we're growing it

Context Q3/4: "What's your primary concern right now?"

1. Will anyone actually want this? (demand/desirability risk)
2. Can users figure out how to use it? (usability risk)
3. Can we actually build and sustain this? (technical/feasibility risk)
4. Will this work as a business? (viability/business model risk)
5. Something external could disrupt our plans (regulatory, market, competitive)

Context Q4/4: "Who is the customer — the person who would use or buy this product? Give me a job title or role, not a company name."

After Q4, agent summarizes:

"Here's what I'm working with:

• Idea: [summary]
• Stage: [stage]
• Primary concern: [concern]
• Customer: [role]

Let's scan for risks. We'll start with internal risks (Product Outcome Metrics and Business Outcome Metrics), then move to external forces (PESTEL)."

Step 1: Internal Risk Scan — Product Outcome Metrics (Desirability + Usability)

Agent introduces:

"First, let's check Product Outcome Metrics — what's in it for the customer? These measure whether the product will be valuable and usable for your customer. I'll present tests and metrics for two dimensions. For each, tell me which 2 are most critical for your product."

Risk Scan Q1/10 — Desirability: "Which of these tests or metrics matter most for validating that [customer role] will actually want [product idea]? Pick 2."

1. Use landing pages to gauge initial interest
2. Test engagement with explainer videos or ads
3. Analyze signup rates from ad campaigns
4. Observe feedback from storyboard demos
5. Measure session times during user interactions
6. Track feature adoption rates over time
7. Other (describe your own)

For each selection, agent asks: "Is this something you need to act on immediately (run the test, close the gap) or start tracking (monitor over time)?"

Risk Scan Q2/10 — Usability: "Which of these matter most for validating that [customer role] can actually use [product idea] without friction? Pick 2."

1. Watch how users interact with wireframes
2. Use Wizard of Oz tests for usability insights
3. Observe clarity issues from paper model feedback
4. Track rage clicks to identify frustration points
5. Measure drop-offs in task completion workflows
6. Analyze time-to-complete tasks for friction points
7. Other (describe your own)

For each selection, agent asks: "Act on immediately or start tracking?"

Step 2: Internal Risk Scan — Business Outcome Metrics (Feasibility + Viability)

Agent introduces:

"Now let's check Business Outcome Metrics — what's in it for our business? These measure whether the business and technology can support this product."

Risk Scan Q3/10 — Feasibility: "Which of these matter most for confirming you can actually build and sustain [product idea]? Pick 2."

1. Measure system uptime and stability
2. Simulate user loads to ensure scalability
3. Track resource use and efficiency during testing
4. Monitor failure rates across environments
5. Evaluate costs via analogous systems
6. Test integration readiness with critical dependencies
7. Other (describe your own)

For each selection, agent asks: "Act on immediately or start tracking?"

Risk Scan Q4/10 — Viability: "Which of these matter most for confirming [product idea] will work as a business? Pick 2."

1. Examine analogous data from similar products
2. Analyze adjacent products for comparable insights
3. Estimate break-even points using simulation models
4. Test price sensitivity through customer research
5. Assess scalability of revenue vs. cost structures
6. Run trend analysis to evaluate long-term market potential
7. Other (describe your own)

For each selection, agent asks: "Act on immediately or start tracking?"

Step 3: External Risk Scan — PESTEL

Agent introduces:

"Now let's scan for external forces — things outside your organization that could reshape, delay, or kill your plans. I'll walk through 6 categories. For each, I'll suggest risks relevant to your product and industry. Tell me which ones matter, and whether to act or watch."

Risk Scan Q5/10 — Political: "Given [product idea] and [customer role], which political forces could affect your plans?"

Starter prompts (agent adapts to context):

• Regulatory compliance requirements (existing or pending)
• Changing political priorities that affect your industry
• Regional policy mandates or jurisdictional differences
• Export restrictions or trade policy shifts

"Which of these (or others) are relevant? For each, act on immediately or start tracking?"

Risk Scan Q6/10 — Economic: "What economic forces could affect [product idea]?"

Starter prompts:

• Economic inequality affecting your customer segment's budget
• Supply chain disruptions impacting delivery or cost
• Funding environment changes (VC, enterprise budgets)
• Skills shortages in your hiring market

"Which are relevant? Act or watch?"

Risk Scan Q7/10 — Social: "What social or cultural shifts could affect adoption of [product idea]?"

Starter prompts:

• Shifting social sentiment toward AI or your product category
• Digital literacy gaps in your target customer population
• Demographic shifts in your target market
• Urbanization or remote-work trends affecting how customers work

"Which are relevant? Act or watch?"

Risk Scan Q8/10 — Technological: "What technology shifts could affect [product idea]?"

Starter prompts:

• Platform dependencies (APIs, cloud providers, AI model providers)
• Evolving industry standards that your product must meet
• Competing technology that could leapfrog your approach
• Spectrum allocations or infrastructure changes (if applicable)

"Which are relevant? Act or watch?"

Risk Scan Q9/10 — Environmental: "What environmental factors could affect [product idea]?"

Starter prompts:

• Energy consumption concerns (especially for AI/compute-heavy products)
• Sustainability requirements from customers or regulators
• Carbon footprint scrutiny from public or investors
• Resource scarcity affecting infrastructure or supply

"Which are relevant? For many products this category has low impact — if so, say 'minimal' and we'll move on. Act or watch for anything relevant?"

Risk Scan Q10/10 — Legal: "What legal risks could affect [product idea]?"

Starter prompts:

• Data privacy regulations (GDPR, state privacy laws, sector-specific rules)
• AI-specific liability or transparency requirements
• IP protection challenges (patents, trade secrets, open-source licensing)
• Employment or contractor classification laws (if product affects work)

"Which are relevant? Act or watch?"

Step 4: Synthesize — Risk Register

Agent compiles all selections into a risk register:

"Here's your de-risking plan for [product idea]:

Act on Immediately

Start Tracking

Summary

• Total risks surfaced: [N]
• Act on immediately: [N] — these are your next steps before committing further
• Start tracking: [N] — set calendar reminders for these review cadences
• Biggest gap: [the dimension with the most unresolved act-now items]"

Agent then offers:

"What would you like to do next?

1. Dig deeper on an act-now item — I'll help you design the specific test or measurement
2. Run a full PESTEL analysis — use the pestel-analysis skill for a deep external-forces dive
3. Pick a validation method — use the pol-probe-advisor to choose the cheapest test for your top risk
4. Export this register — I'll format it as a Markdown document you can share with your team"

Examples

Example: AI-Powered Clinical Trial Matching Tool

Context:

• Idea: AI tool that matches patients to clinical trials based on EHR data
• Stage: Concept
• Primary concern: External regulatory risk
• Customer: Clinical operations lead

Internal Risk Scan Results:

Product Outcome Metrics:

• Desirability: Observe feedback from storyboard demos (ACT) + Track feature adoption rates (WATCH)
• Usability: Measure drop-offs in task completion workflows (WATCH) + Analyze time-to-complete for friction (WATCH)

Business Outcome Metrics:

• Feasibility: Test integration readiness with EHR dependencies (ACT) + Monitor failure rates across environments (WATCH)
• Viability: Examine analogous data from similar products (ACT) + Assess scalability of revenue vs. cost (WATCH)

External Risk Scan Results:

• Political: FDA digital health guidance evolving (WATCH — quarterly)
• Economic: Hospital IT budget pressure post-pandemic (ACT — validate willingness to pay now)
• Social: Clinician trust in AI-driven patient matching (ACT — run trust interviews)
• Technological: EHR platform API stability and access policies (ACT — test integration this sprint)
• Environmental: Minimal impact
• Legal: HIPAA compliance for AI processing of PHI (ACT — legal review before prototype)

Risk Register Output:

Act on immediately (5 items):

1. Desirability — storyboard demos with 5 clinical ops leads
2. Feasibility — EHR integration spike (test API access and data format)
3. Viability — analogous product benchmarking (what do similar tools charge?)
4. Economic — validate budget authority with 3 hospital IT buyers
5. Legal — HIPAA compliance review for AI processing of PHI

Start tracking (5 items):

1. Desirability — feature adoption rates (post-launch metric)
2. Usability — task completion drop-offs and time-to-complete (post-prototype)
3. Feasibility — failure rates across environments (post-build)
4. Viability — revenue vs. cost scalability (post-pilot)
5. Political — FDA digital health guidance (quarterly review)

Biggest gap: Legal/compliance — HIPAA review must happen before any prototype touches real patient data.

Common Pitfalls

1. Generic Risk Lists

Failure Mode: "Desirability is a risk. Feasibility is a risk. Regulation is a risk." No specifics, no connection to the actual product.

Consequence: Risk register that applies to every product — and therefore helps none.

Fix: Every risk must name the specific test or signal for this product and this customer. "Gauge initial interest" is generic. "Run a 1-week landing page targeting clinical ops leads with EHR integration messaging" is specific.

2. All Act, No Watch

Failure Mode: Every risk is labeled "act on immediately" because everything feels urgent when you're looking at it.

Consequence: No prioritization. Team tries to address 15 risks simultaneously and makes progress on none.

Fix: Force a split. If you have 10 risks, no more than 4-5 should be "act now." The rest are real risks — they just aren't today's problem. Set a review cadence and move on.

3. Internal-Only or External-Only

Failure Mode: PM runs a thorough DUFV scan but never checks PESTEL (or vice versa). "We validated demand and feasibility — ship it!" Then a regulation kills the product 6 months later.

Consequence: Blind spots in exactly the dimensions you didn't scan.

Fix: This skill exists precisely to prevent this. Both lenses, every time. If an external category has low impact (common for Environmental in SaaS), say "minimal" and move on — but don't skip the question.

4. Analysis Without Next Steps

Failure Mode: Beautiful risk register, no recommended first step. Sits in a Google Doc. Nobody acts on it.

Consequence: Risk theater — the appearance of diligence without the substance.

Fix: Every "act now" item must have a concrete next step: who does what, by when, and how you'll know the risk is addressed. "Run a test" is not a next step. "Launch a 1-week landing page with email capture targeting [customer role], measure signup rate, and decide go/no-go by [date]" is.

References

Related Skills

• pestel-analysis (Component) — Deep-dive PESTEL template with data-driven analysis across all 6 categories; use after this skill surfaces external risks worth investigating further
• pol-probe-advisor (Interactive) — Choose the cheapest validation method for your highest-priority act-now risk
• pol-probe (Component) — Document a Proof of Life probe for a specific risk
• recommendation-canvas (Component) — Structure a go/no-go recommendation using risk register findings
• problem-framing-canvas (Interactive) — Frame the problem before scanning for risks

External Frameworks

• Marty Cagan — Inspired (2018): The four product risks (Value, Usability, Feasibility, Viability) as the foundation for product discovery
• Francis Joseph Aguilar — Scanning the Business Environment (1967): Origin of PEST analysis, extended to PESTEL

Dean's Work

• Productside AI Product Management course, Unit 07: "De-Risking Your AI Bets — Unpacking Unknowns and Assumptions"
• PESTel Risk Planning activity (external risks with act/watch triage)
• Measuring What Matters activity (internal risks with DUFV quadrant selection)

Purpose

Guide PMs and Directors through the specific challenges of the PM-to-Director transition using adaptive questions and targeted coaching. Diagnoses where you are in the journey and delivers practical, war-story-backed guidance calibrated to your situation — not generic leadership advice.

This is not a readiness checklist. It's a coaching conversation that names what's actually hard, why it's hard, and what to do about it.

Input

Works best with: Where you are in the PM-to-Director journey: preparing for the jump, interviewing, newly landed, or recalibrating after a rough start. Also useful: Your org context (team size, who you'd report to) and the specific friction you're feeling.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The advisor opens by asking which of the four situations describes you, then branches.

Example invocation: I'm 60 days into a Director role and my old peers now report to me — it's awkward and I'm still doing PM work. Coach me.

Key Concepts

The Four Transition Situations

The PM → Director transition looks different depending on where you are:

1. Preparing to make the leap — Still a PM, actively developing toward the role
2. Interviewing for Director roles — In an active internal or external search
3. Newly landed — Recently promoted or hired as Director (first 6 months)
4. Recalibrating — Been a Director for a while; something isn't working

Each situation has distinct coaching priorities. The biggest mistake is applying "newly landed" advice to someone who's been in the role for two years, or "preparing" advice to someone mid-interview process.

The Underlying Model

This skill draws directly on the Altitude & Horizon Framework — see skills/altitude-horizon-framework/SKILL.md for the full mental model. Core concepts used here:

• Altitude (scope) and Horizon (time) as the two axes that shift
• The Waiter vs. Restaurant Operator distinction
• Four transition zones: Thinking Altitude, Persona Shift, Hero Syndrome Recovery, Direction Creation
• Named failure modes: Hero Syndrome, Allergic to Process, People-Pleaser Leadership, Instant Gratification Trap

Facilitation Source of Truth

Use workshop-facilitation as the default interaction protocol for this skill.

It defines:

• Session heads-up + entry mode (Guided, Context dump, Best guess)
• One-question turns with plain-language prompts
• Progress labels (e.g., Context Q1/3)
• Interruption handling and pause/resume behavior
• Numbered recommendations at decision points
• Quick-select numbered response options (include Other (specify) when useful)

This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.

Application

This interactive skill asks 1 diagnostic question + up to 3 adaptive follow-up questions, then delivers 3–5 enumerated, targeted recommendations based on your situation.

Step 0: Session Start

Agent says:

Before we start, you can choose how to run this session:

1. Guided — I'll ask questions one at a time and build recommendations from your answers (recommended for most situations)
2. Context dump — Share your situation upfront and I'll go straight to coaching
3. Best guess — Tell me nothing; I'll give you the highest-value advice for the most common transition situation (newly landed Director, 0–3 months in)

Question 1: Where Are You?

Agent asks (Q1/3):

"Where are you in the PM-to-Director journey?"

1. Preparing to make the leap — Still in a PM role, actively building toward Director
2. Interviewing for Director roles — In an active internal promotion or external job search
3. Newly landed — Recently promoted or hired as Director (roughly first 6 months)
4. Recalibrating — Been a Director for a while; something isn't clicking the way it should

Or describe your situation directly.

Branch 1: Preparing to Make the Leap

Question 1B (Q2/3):

Agent asks:

"What's the gap you most want to close before making the move?"

1. Thinking altitude — I still default to solving customer problems directly instead of designing systems
2. Stakeholder navigation — I struggle with organizational politics, executive dynamics, or cross-functional influence
3. Strategic narrative — I can't yet connect my work to company strategy convincingly in conversations with leadership
4. Direction creation — I wait for clarity from above rather than creating it for my team

Or describe your main development area.

Question 1C (Q3/3):

Agent asks:

"What's your runway — how close are you to making the move?"

1. 6+ months out — Building deliberately; no immediate pressure
2. 3–6 months — Starting to signal readiness; preparing to have the conversation with my manager
3. Actively applying — In motion now, interviewing or expecting to be soon

Or describe your timeline.

Branch 1 Recommendations:

Agent delivers (based on 1B + 1C):

Example output for 1B = Strategic narrative + 1C = 3–6 months:

"Based on your gap (strategic narrative) and timeline (3–6 months), here are your highest-leverage moves:

1. Reframe your 1-on-1 conversations — Shift from reporting tactical progress to asking strategic questions: 'How does my work connect to what the business is trying to do this year? What's the organizational context I'm not seeing?' This builds strategic thinking muscle and signals readiness to your manager simultaneously.

2. Build your cascade habit now — Pick one company priority from the last QBR or all-hands and write a one-page translation: how does your team's work connect? Share it informally with your manager. You don't need permission to practice this skill.

3. Audit your Hero Syndrome — Track the next two weeks: how often do you jump in to solve problems that others could solve with your coaching instead? The ratio of rescuing to developing is a leading indicator of Director readiness.

4. Don't over-index on Director thinking — The most common mistake while still a PM: spending too much energy on altitude/horizon thinking at the expense of serving your current role. Play your PM role fully. Director readiness will show in the quality of your work, not in performing above your level.

Choose a number to go deeper on any of these, or ask about something specific."

Branch 2: Interviewing for Director Roles

Question 2B (Q2/3):

Agent asks:

"What's your interview context?"

1. Internal promotion — Being considered within my current company
2. External search — Interviewing at new companies
3. Both — Running an internal and external process simultaneously

Question 2C (Q3/3):

Agent asks:

"What do you most want to work on for interview prep?"

1. Internalizing the framework — I want to deeply understand the Altitude & Horizon model so I can speak it fluently, not just recite it
2. Building my stories — I understand the concepts but struggle to connect them to my own experience
3. Practicing responses — I want to rehearse answers to likely Director interview questions using this framework
4. Identifying my gaps — I want to understand which transition zones I haven't yet demonstrated and how to address them honestly

Or describe what you most need.

Branch 2 Recommendations:

Agent delivers (based on 2B + 2C):

Example output for 2B = External + 2C = Practicing responses:

"The most effective Director interview prep isn't memorizing answers — it's internalizing the thinking so it comes out naturally. These two skills are your prep tools:

1. Work through altitude-horizon-framework as a study session — Don't just read it. After each section, pause and answer: 'Where have I operated at this altitude or horizon in my own work? What's my story here?' The Four Transition Zones and Named Failure Modes map almost directly to common Director interview questions.

2. Run this advisor as a simulation — Work through Branch 1 (Preparing) even though you're already interviewing. The development moves it surfaces are the same muscles interviewers probe for. If you can't do the exercise, you haven't yet built the answer.

3. Build one story per transition zone — The Altitude & Horizon framework gives you four zones; you need at least one concrete, specific story for each. Use the zone names as the structure: 'Here's a time I had to shift my thinking altitude... here's what triggered the persona shift... here's when I had to create direction rather than wait for it.' Interviewers recognize this kind of self-awareness immediately.

4. Reframe your PM wins in Director language — You almost certainly have the experience; it's framed wrong. Don't open with 'I shipped X.' Open with: 'The strategic question my team was facing was [X]. Here's how I thought about the portfolio tradeoff.' Use the feature or outcome as evidence, not as the headline.

5. Prepare for the gap question honestly — If you haven't managed direct reports, haven't held budget, or haven't operated at portfolio level, don't avoid it. The strongest answer is: 'Here's the gap, here's how I've been developing toward it, and here's what I'd focus on learning in the first 90 days.' Honesty with a plan lands significantly better than avoidance.

Choose a number to go deeper, or describe what you want to practice."

Branch 3: Newly Landed

Question 3B (Q2/3):

Agent asks:

"What's your most pressing challenge right now?"

1. Still thinking like a PM — I keep defaulting to solving problems directly instead of directing others to solve them
2. Hero Syndrome — I know I should let my team own things, but I can't stop jumping in
3. Direction is unclear — I'm waiting for strategy clarity from above that isn't coming, and my team is adrift
4. Stakeholder shift — I'm still relating to the wrong people — too much time with engineers, not enough with peers and executives

Or describe what's most pressing.

Question 3C (Q3/3):

Agent asks:

"What's your team situation?"

1. Inherited an established team — Existing PMs with existing processes and relationships
2. Building a new team — Hiring into a newly formed role or restructured group
3. Mixed — Some experienced team members, some new, some in transition

Branch 3 Recommendations:

Agent delivers (based on 3B + 3C):

Example output for 3B = Direction is unclear + 3C = Inherited team:

"An inherited team without clear direction is the most common Director failure mode in the first 90 days. Here's what to do:

1. Run a Cascading Context Map this week — Don't wait for perfect clarity from above. Take the most recent company priority you heard (QBR, all-hands, your manager's stated objectives) and translate it down: what does it mean for your business unit, your portfolio, your team's specific accountabilities? One page. Share it. Revise it when you learn more.

2. Reframe your 1-on-1s immediately — Most inherited teams will use 1-on-1s to brief you on feature status and tactical priorities. Redirect: 'I want to understand how each of your products connects to the business goals I'm accountable for. Help me see the through-line.' This establishes altitude from day one.

3. Name the ambiguity explicitly to your team — Don't pretend you have clarity you don't have. 'I'm still building my picture of where leadership wants us to go. Here's my best current translation. I'll update it in two weeks.' Teams would rather have honest partial context than silence.

4. Resist the urge to reorganize before you understand — With an inherited team, the instinct is to reorganize processes and structures to reflect how you think. Wait 60–90 days. Understand what's working and why before you change it. The monkeys in the room aren't always wrong — some of them built the place.

Choose a number to go deeper, or describe what's most blocking you."

Branch 4: Recalibrating

Question 4B (Q2/3):

Agent asks:

"What's the core friction? Where does the role feel most broken right now?"

1. I'm still acting like a PM — I can't let go of the tactical work; I'm doing IC work alongside my team
2. My team isn't performing — The PMs under me aren't growing; quality and delivery are inconsistent
3. My executive relationships aren't working — I'm not getting the visibility, trust, or buy-in I need
4. I don't have portfolio clarity — My product portfolio lacks coherent strategy; I'm managing a collection of roadmaps, not a system

Or describe what's not working.

Question 4C (Q3/3):

Agent asks:

"How long have you been in this Director role?"

1. Under 1 year — Still in the transition period
2. 1–2 years — Past the transition; this feels like a persistent pattern
3. 2+ years — Established in the role; this has become a structural issue

Branch 4 Recommendations:

Agent delivers (based on 4B + 4C):

Example output for 4B = Still acting like a PM + 4C = 1–2 years:

"A Director who is still doing IC work after 12+ months isn't transitioning — they've settled into an unsustainable hybrid role. Here's the diagnosis and the fix:

1. Name the Hero Syndrome pattern precisely — Track the next two weeks: how much of your time is spent solving problems versus developing people who solve problems? Most Directors in this pattern are at 60–70% IC work. The target is closer to 20%.

2. Identify what's keeping you in it — There are usually three causes: (a) you trust your own judgment more than your team's, (b) the team hasn't been developed enough to own things without you, (c) you're getting reward signals (praise, visibility) that reinforce the IC behavior. Which one is it?

3. Create a deliberate handoff for your top 3 IC activities — List the three things you do most often that a PM should own. For each: write down what 'done well' looks like, have a conversation with the PM about owning it, and then don't rescue when they struggle. Coaching through failure is the work.

4. Change the reward loop — The pat-on-the-back you got as a PM doesn't come as a Director. Director success is quieter and more delayed. Find the new signal: a PM who ships a difficult stakeholder conversation alone, a team that creates its own context cascade, a portfolio decision you made that held up under pressure. Start noticing those wins.

5. If this pattern is entrenched at 1–2 years, consider whether the role is the right fit — Some people are happier and more effective as senior ICs or Principal PMs. That's not failure — it's self-knowledge. The IC path is legitimate; the mismatch is staying in a Director role while operating as a PM.

Choose a number to go deeper, or tell me what you want to work on next."

Examples

See examples/conversation-flow.md for a full end-to-end interaction, including context intake, branch selection, and final recommendations.

Example: Newly Landed Director, Unclear Direction

Q1: "3 — Newly landed" Q2: "3 — Direction is unclear" Q3: "1 — Inherited an established team"

Agent output: Cascading Context Map instructions, advice on reframing 1-on-1s, naming ambiguity explicitly, and resisting premature reorganization. Links to altitude-horizon-framework for the full cascade template.

Example: PM Preparing, Strategic Narrative Gap

Q1: "1 — Preparing to make the leap" Q2: "3 — Strategic narrative" Q3: "2 — 3–6 months"

Agent output: 1-on-1 reframe tactics, cascade habit-building exercise, Hero Syndrome audit, and warning against over-indexing on Director thinking while still in PM role.

Common Pitfalls

Pitfall 1: Treating Preparation Like a Checklist

Symptom: Asking "what do I need to do to get promoted?" and working through items like tasks

Consequence: You optimize for appearances of readiness rather than building the actual muscles. Interviewers and managers can tell the difference.

Fix: Use this advisor to identify the one or two specific behavior changes that matter most for your situation, not a comprehensive development program.

Pitfall 2: Misidentifying Your Situation

Symptom: Selecting "preparing" when you're actually in an active interview process, or selecting "newly landed" when you've been in the role 18 months

Consequence: You get coaching calibrated to the wrong situation.

Fix: Be honest about where you actually are, not where you'd like to be.

Pitfall 3: Looking for Permission

Symptom: Asking this advisor whether you're "ready" for the transition

Consequence: There's no readiness test. The transition is a decision, not a graduation.

Fix: Use this skill to identify what will be hardest for you specifically and how to address it — not to get a pass/fail verdict.

References

Related Skills

• skills/altitude-horizon-framework/SKILL.md — The mental model this skill coaches on; includes the Cascading Context Map template
• skills/workshop-facilitation/SKILL.md — Facilitation protocol for this interactive skill

Future Skills

• director-to-vp-cpo-advisor (planned) — Coaches the next transition: Director to VP or CPO of Product

Source Material

• The Product Porch, Episode 42: From Product Manager to Director: How to Make the Shift (Part 1) — Todd Blaquiere, Ryan Cantwell, Joe Ghali (January 2026)

External Frameworks

• Marty Cagan, Empowered — Organizational dynamics and role clarity
• Julie Zhuo, The Making of a Manager — IC-to-manager transition
• Michael Watkins, The First 90 Days — Structured approach to leadership transitions

Purpose

Guide product managers through preparing for customer discovery interviews by asking adaptive questions about research goals, customer segments, constraints, and methodologies. Use this to design effective interview plans, craft targeted questions, avoid common biases, and maximize learning from limited customer access—ensuring discovery interviews yield actionable insights rather than confirmation bias or surface-level feedback.

This is not a script generator—it's a strategic prep process that outputs a tailored interview plan with methodology, question framework, and success criteria.

Input

Works best with: Your research goal — what you need to learn from customers. Also useful: Customer segment, access constraints (how many interviews, by when), and any hypotheses you're carrying in.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The skill opens by asking your primary goal for the interviews, then narrows with follow-ups.

Example invocation: Prep interviews to understand why enterprise customers churn after 6 months — I can get 5 interviews in 2 weeks.

Key Concepts

The Discovery Interview Prep Flow

An interactive process that:

1. Gathers product/problem context (marketing materials, assumptions)
2. Defines research goals (what you're trying to learn)
3. Identifies target customer segment and access constraints
4. Recommends interview methodology (Jobs-to-be-Done, problem validation, switch interviews, etc.)
5. Generates interview framework with questions, biases to avoid, and success metrics

Why This Works

• Goal-driven: Aligns interview approach to what you need to learn
• Adaptive: Adjusts methodology based on product stage (idea vs. existing product) and access constraints
• Bias-aware: Highlights common pitfalls (leading questions, confirmation bias, solution-first thinking)
• Actionable: Outputs interview guide ready to use

Anti-Patterns (What This Is NOT)

• Not a user testing script: Discovery = learning problems; testing = validating solutions
• Not a sales demo: Don't pitch—listen and learn
• Not surveys at scale: Deep qualitative interviews (5-10 people), not broad surveys (100+ people)

When to Use This

• Starting product discovery (validating problem space)
• Repositioning an existing product (understanding new market)
• Investigating churn or drop-off (retention interviews)
• Evaluating feature ideas before building
• Preparing for customer development sprints

When NOT to Use This

• User testing a prototype (use usability testing frameworks instead)
• Quantitative research at scale (use surveys, analytics)
• When you already know the problem (move to solution validation)

Facilitation Source of Truth

Use workshop-facilitation as the default interaction protocol for this skill.

It defines:

• session heads-up + entry mode (Guided, Context dump, Best guess)
• one-question turns with plain-language prompts
• progress labels (for example, Context Qx/8 and Scoring Qx/5)
• interruption handling and pause/resume behavior
• numbered recommendations at decision points
• quick-select numbered response options for regular questions (include Other (specify) when useful)

This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.

Application

This interactive skill asks up to 4 adaptive questions, offering 3-4 enumerated options at each step.

Step 0: Gather Context (Before Questions)

Agent suggests:

Before we design your interview plan, let's gather context:

For Your Own Product (Existing or Planned):

• Problem hypothesis or product concept description
• Target customer segment (if known)
• Existing research (support tickets, churn data, user feedback)
• Product website or positioning materials
• Key assumptions you're trying to validate

For Investigating an Existing Problem:

• Customer complaints, support tickets, or churn reasons
• Hypotheses about why customers leave or struggle
• Competitive alternatives customers switch to

If Exploring a New Problem Space:

• Find similar products or adjacent solutions
• Copy competitor materials, customer reviews (G2, Capterra), or community discussions (Reddit, forums)
• We'll use these to frame hypotheses

You can paste this content directly, or we can proceed with a brief description.

Question 1: Research Goal

Agent asks: "What's the primary goal of these discovery interviews? (What do you need to learn?)"

Offer 4 enumerated options:

1. Problem validation — "Confirm that a problem exists and is painful enough to solve" (for new product ideas)
2. Jobs-to-be-Done discovery — "Understand what customers are trying to accomplish and why current solutions fail" (for product strategy)
3. Retention/churn investigation — "Figure out why customers leave or don't activate" (for existing products with churn issues)
4. Feature prioritization — "Validate which problems/features matter most to customers" (for roadmap planning)

Or describe your own research goal (be specific: what question are you trying to answer?).

User response: [Selection or custom]

Question 2: Target Customer Segment

Agent asks: "Who are you interviewing? (Be as specific as possible.)"

Offer 4 enumerated options (adapted based on Q1):

Example (if Q1 = Problem validation):

1. People who experience the problem regularly — E.g., "Small business owners who manually process invoices weekly" (high pain frequency)
2. People who've tried to solve it — E.g., "Users who've tried 2+ competing solutions and churned" (understand failures)
3. People in the target segment (regardless of problem awareness) — E.g., "All freelancers, even if they don't realize invoicing is a problem" (uncover latent needs)
4. People who've recently experienced the problem — E.g., "Customers who churned in the last 30 days" (fresh memory)

Or describe your specific target segment (role, company size, behaviors, demographics).

Adaptation tip: Use personas or customer segments from provided materials.

User response: [Selection or custom]

Question 3: Constraints

Agent asks: "What constraints are you working with for these interviews?"

Offer 4 enumerated options:

1. Limited access — "Can only interview 5-10 customers, need results in 2 weeks" (common for startups or fast timelines)
2. Existing customer base — "Have 100+ active customers, can recruit easily" (mature product advantage)
3. Cold outreach required — "No existing customers; need to recruit from scratch via LinkedIn, ads, or communities" (new product challenge)
4. Internal stakeholders only — "Can interview sales/support teams who talk to customers daily" (proxy research, less ideal but pragmatic)

Or describe your specific constraints (budget, time, access, team capacity).

User response: [Selection or custom]

Question 4: Interview Methodology

Agent asks: "Based on your goal ([Q1]), target segment ([Q2]), and constraints ([Q3]), here are recommended interview methodologies:"

Offer 3-4 enumerated options (context-aware based on Q1-Q3):

Example (if Q1 = Problem validation, Q2 = People who experience problem regularly, Q3 = Limited access):

1. Problem validation interviews (Mom Test style) — Ask about past behavior, not hypotheticals. Focus on: "Tell me about the last time you [experienced the problem]. What did you try? What happened?" (Best for: Validating if problem is real and painful)

2. Jobs-to-be-Done (JTBD) interviews — Focus on what customers are trying to accomplish, not what they want. Ask: "What were you trying to get done? What alternatives did you consider? What made you choose X?" (Best for: Understanding motivations and switching behavior)

3. Switch interviews — Interview customers who recently switched from a competitor or alternative. Ask: "What prompted you to look for a new solution? What was the 'push' away from the old tool? What 'pulled' you to try ours?" (Best for: Understanding competitive positioning and unmet needs)

4. Timeline/journey mapping interviews — Walk through their entire experience chronologically. Ask: "Walk me through the first time you encountered this problem. What happened next? How did you try to solve it?" (Best for: Uncovering full context and pain points)

Choose a number, combine approaches (e.g., '1 & 2'), or describe your own methodology.

Adaptation examples:

• If Q1 = Retention/churn → Prioritize "Exit interviews" or "Switch interviews (away from your product)"
• If Q1 = Feature prioritization → Prioritize "Opportunity solution tree interviews" or "Kano model interviews"
• If Q3 = Internal stakeholders only → Add caveat: "Proxy research (talking to sales/support) is better than nothing, but validate with real customers ASAP"

User response: [Selection or custom]

Output: Generate Interview Plan

After collecting responses, the agent generates a tailored interview plan:

# Discovery Interview Plan

**Research Goal:** [From Q1]
**Target Segment:** [From Q2]
**Constraints:** [From Q3]
**Methodology:** [From Q4]

---

## Interview Framework

### Opening (5 minutes)
- **Build rapport:** "Thanks for taking the time. I'm [name], and I'm researching [problem space]. This isn't a sales call—I'm here to learn from your experience."
- **Set expectations:** "I'll ask about your experiences with [topic]. There are no right answers. Feel free to be honest—critical feedback is most helpful."
- **Get consent:** "Is it okay if I take notes / record this conversation?"

---

### Core Questions (30-40 minutes)

**Based on your methodology ([Q4]), here are suggested questions:**

#### [Methodology Name] Questions:

1. **[Question 1]** — [Rationale for asking this]
   - **Follow-up:** [Dig deeper with...]
   - **Avoid:** [Don't ask leading version like...]

2. **[Question 2]** — [Rationale]
   - **Follow-up:** [...]
   - **Avoid:** [...]

3. **[Question 3]** — [Rationale]
   - **Follow-up:** [...]
   - **Avoid:** [...]

4. **[Question 4]** — [Rationale]
   - **Follow-up:** [...]
   - **Avoid:** [...]

5. **[Question 5]** — [Rationale]
   - **Follow-up:** [...]
   - **Avoid:** [...]

**Example (if Methodology = Problem validation - Mom Test style):**

1. **"Tell me about the last time you [experienced this problem]."** — Gets specific, recent behavior (not hypothetical)
   - **Follow-up:** "What were you trying to accomplish? What made it hard? What did you try?"
   - **Avoid:** "Would you use a tool that solves this?" (leading, hypothetical)

2. **"How do you currently handle [this problem]?"** — Reveals workarounds, alternatives, pain intensity
   - **Follow-up:** "How much time/money does that take? What's frustrating about it?"
   - **Avoid:** "Don't you think that's inefficient?" (leading)

3. **"Can you walk me through what you did step-by-step?"** — Uncovers details, edge cases, context
   - **Follow-up:** "What happened next? Where did you get stuck?"
   - **Avoid:** "Was it hard?" (yes/no question, not useful)

4. **"Have you tried other solutions for this?"** — Reveals competitive landscape, unmet needs
   - **Follow-up:** "What did you like/dislike? Why did you stop using it?"
   - **Avoid:** "Would you pay for a better solution?" (hypothetical)

5. **"If you had a magic wand, what would change?"** — Opens space for ideal outcomes (but treat with skepticism—focus on past behavior, not wishes)
   - **Follow-up:** "Why does that matter to you? What would that enable?"
   - **Avoid:** Taking feature requests literally

---

### Closing (5 minutes)
- **Summarize:** "Just to recap, I heard that [key insights]. Did I get that right?"
- **Ask for referrals:** "Do you know anyone else who experiences this problem? Could you introduce me?"
- **Thank them:** "This was incredibly helpful. I really appreciate your time."

---

## Biases to Avoid

1. **Confirmation bias:** Don't ask "Don't you think X is a problem?" → Ask "Tell me about your experience with X."
2. **Leading questions:** Don't ask "Would you use this?" → Ask "What have you tried? Why did it work/fail?"
3. **Hypothetical questions:** Don't ask "If we built Y, would you pay?" → Ask "What do you currently pay for? Why?"
4. **Pitching disguised as research:** Don't say "We're building Z to solve X" → Say "I'm researching X. Tell me about your experience."
5. **Yes/no questions:** Don't ask "Is invoicing hard?" → Ask "Walk me through your invoicing process."

---

## Success Criteria

You'll know these interviews are successful if:

✅ **You hear specific stories, not generic complaints** — "Last Tuesday, I spent 3 hours..." vs. "Invoicing is annoying"
✅ **You uncover past behavior, not hypothetical wishes** — "I tried Zapier but quit after 2 weeks" vs. "I'd probably use automation"
✅ **You identify patterns across 3+ interviews** — Same pain points emerge independently
✅ **You're surprised by something** — If everything confirms your assumptions, you're asking leading questions
✅ **You can quote customers verbatim** — Actual language = authentic insights

---

## Interview Logistics

**Recruiting:**
- [Based on Q3 constraints, suggest recruitment channels]
- **Example (if Q3 = Limited access):** "Reach out to 20-30 people to get 5-10 interviews (33% response rate is typical)"
- **Example (if Q3 = Existing customers):** "Email 50 customers with $50 Amazon gift card incentive"

**Scheduling:**
- 45-60 minutes per interview (30-40 min conversation + buffer)
- Record if possible (with consent), or take detailed notes
- Schedule 2-3 per day max (you need time to synthesize)

**Synthesis:**
- After each interview, write key insights immediately (memory fades fast)
- After 5 interviews, look for patterns (common pains, jobs, workarounds)
- Use `problem-statement.md` to frame findings

---

**Ready to start recruiting and interviewing? Let me know if you'd like to refine any part of this plan.**

Examples

Example 1: Good Discovery Interview Prep (Problem Validation)

Step 0 - Context: User shares hypothesis: "Freelancers waste time chasing late payments manually."

Q1 Response: "Problem validation — Confirm that late payment follow-ups are painful enough to solve"

Q2 Response: "People who experience the problem regularly — Freelancers who invoice 5+ clients monthly"

Q3 Response: "Cold outreach required — No existing customers; need to recruit via LinkedIn, Reddit, freelancer communities"

Q4 Response: "Problem validation interviews (Mom Test style) — Focus on past behavior, not hypotheticals"

Generated Plan: Includes 5 Mom Test-style questions (last time you chased a late payment, how do you currently handle it, what have you tried, etc.), biases to avoid (leading questions, hypotheticals), and success criteria (specific stories, past behavior, patterns across 3+ interviews).

Why this works:

• Goal is clear (validate if problem is real)
• Segment is specific (freelancers with 5+ clients/month)
• Methodology matches goal (Mom Test for validation)
• Questions focus on past behavior, not wishes
• Success criteria are measurable

Common Pitfalls

Pitfall 1: Asking What Customers Want

Symptom: "What features do you want us to build?"

Consequence: You get feature requests, not problems. Customers don't know solutions.

Fix: Ask about past behavior: "Tell me about the last time you struggled with X."

Pitfall 2: Pitching Instead of Listening

Symptom: Spending 20 minutes explaining your product idea

Consequence: Customer feels obligated to be nice. No honest feedback.

Fix: Don't mention your solution until the last 5 minutes (if at all). Focus on their problems.

Pitfall 3: Interviewing the Wrong People

Symptom: Interviewing friends, family, or people who don't experience the problem

Consequence: Polite feedback, not real insights.

Fix: Interview people who experience the problem regularly and recently.

Pitfall 4: Stopping at 1-2 Interviews

Symptom: "We talked to 2 people, they liked it, let's build!"

Consequence: Small sample = confirmation bias.

Fix: Interview 5-10 people minimum. Look for patterns, not one-off feedback.

Pitfall 5: Not Recording Insights

Symptom: Relying on memory after interviews

Consequence: Lose details, misremember quotes, can't spot patterns.

Fix: Record (with consent) or take detailed notes. Synthesize immediately after each interview.

References

Related Skills

• problem-statement.md — Use interview insights to frame problem statement
• proto-persona.md — Define interview target segment
• jobs-to-be-done.md — JTBD methodology for interviews

External Frameworks

• Rob Fitzpatrick, The Mom Test (2013) — How to ask good questions without biasing answers
• Clayton Christensen, Jobs to Be Done — Interview methodology for understanding motivations
• Teresa Torres, Continuous Discovery Habits (2021) — Opportunity solution tree interviews

Dean's Work

• Problem Framing Canvas (synthesizes interview findings)

Skill type: Interactive Suggested filename: discovery-interview-prep.md Suggested placement: /skills/interactive/ Dependencies: Uses problem-statement.md, proto-persona.md, jobs-to-be-done.md

Purpose

Guide product managers through a complete discovery cycle—from initial problem hypothesis to validated solution—by orchestrating problem framing, customer interviews, synthesis, and experimentation skills into a structured process. Use this to systematically explore problem spaces, validate assumptions, and build confidence before committing to full development—avoiding "build it and they will come" syndrome and ensuring you're solving real customer problems.

This is not a one-time research project—it's a continuous discovery practice that runs in parallel with delivery, typically 1-2 discovery cycles per quarter.

Input

Works best with: Your starting problem hypothesis — even a rough one. Also useful: Prior research, customer access, timeline, and what decision the discovery must inform.

Anything supplied with the invocation itself — text after the skill name, a pasted context dump, or an appended ARGUMENTS: line — counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The workflow starts at problem framing and helps you construct the hypothesis first.

Example invocation: Run discovery on this hypothesis: SMB admins abandon onboarding because the data-import step requires IT help they don't have.