メニュー
Identify what to measure, test, or track to de-risk a product or AI idea. Use when stress-testing an idea across internal (DUFV) and external (PESTEL) dimensions.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/deanpeters/Product-Manager-Skills --skill derisk-measurement-advisorコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
SOC 職業分類に基づく
SKILL.md を表示中
| name | derisk-measurement-advisor |
| argument-hint | [product or AI idea] |
| description | Identify what to measure, test, or track to de-risk a product or AI idea. Use when stress-testing an idea across internal (DUFV) and external (PESTEL) dimensions. |
| intent | Guide product managers through a structured risk scan across **10 dimensions** — 4 internal (Desirability, Usability, Feasibility, Viability) and 6 external (Political, Economic, Social, Technological, Environmental, Legal) — to identify the most important things to measure, test, or track before committing to a product or AI idea. Each surfaced risk gets triaged into **act on immediately** or **start tracking**, producing a prioritized risk register with concrete next steps. |
| type | interactive |
| best_for | ["Stress-testing an AI or product idea before putting it on the roadmap","Identifying the cheapest tests to run before committing teams and budget","Scanning for external forces that could blindside your product plans","Building a risk register with clear act-now vs. watch-and-track priorities"] |
| scenarios | ["I have an AI product idea. What should I measure before we commit?","Help me figure out what risks to act on now vs. track over time","We're about to put this on the roadmap. What could go wrong internally and externally?","Is this idea worth the squeeze? What should we test first?"] |
| estimated_time | 15-20 min |
| theme | validation-experiments |
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?
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.
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:
External risks ask whether the world will let it work:
The single triage question that applies to both sides: Do I act on this now, or start tracking it?
Adapted from Marty Cagan's four product risks, split into two measurement categories:
| Category | Dimension | Core Question |
|---|---|---|
| Product Outcome Metrics | Desirability | Will customers value it enough to buy it? |
| Product Outcome Metrics | Usability | Will customers figure out how to use it? |
| Business Outcome Metrics | Feasibility | Can we build it? Can we sustain it at scale? |
| Business Outcome Metrics | Viability | Will it work as a viable business? |
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).
Six categories of forces outside your organization that can reshape, delay, or kill your product plans:
| Category | What It Covers | Example Risk |
|---|---|---|
| Political | Government policies, trade regulations, political priorities | New AI safety rules for state contractors |
| Economic | Market conditions, funding, inequality, supply chains | VC AI funding dries up — focus shifts to profitability |
| Social | Demographics, sentiment, digital literacy, cultural shifts | AI accused of bias — can you prove fairness? |
| Technological | Platform dependencies, standards, competing tech | New AI model outdates yours — time for rapid innovation? |
| Environmental | Energy consumption, sustainability, carbon footprint | AI criticized for carbon footprint — how to go green? |
| Legal | Data privacy, compliance, IP, evolving regulations | AI deepfakes lawsuit crossfire — can you protect your product? |
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.
Every surfaced risk gets one of two labels:
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.
Use this when:
Don't use this when:
problem-framing-canvas first)pestel-analysis component skill)pol-probe-advisor)Use workshop-facilitation as the default interaction protocol for this skill.
It defines:
Other (specify) when useful)This file defines the domain-specific assessment content. If there is a conflict, follow this file's domain logic.
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.
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?
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?"
Context Q3/4: "What's your primary concern right now?"
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:
Let's scan for risks. We'll start with internal risks (Product Outcome Metrics and Business Outcome Metrics), then move to external forces (PESTEL)."
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."
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."
For each selection, agent asks: "Act on immediately or start tracking?"
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."
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."
For each selection, agent asks: "Act on immediately or start tracking?"
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):
"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:
"Which are relevant? Act or watch?"
Risk Scan Q7/10 — Social: "What social or cultural shifts could affect adoption of [product idea]?"
Starter prompts:
"Which are relevant? Act or watch?"
Risk Scan Q8/10 — Technological: "What technology shifts could affect [product idea]?"
Starter prompts:
"Which are relevant? Act or watch?"
Risk Scan Q9/10 — Environmental: "What environmental factors could affect [product idea]?"
Starter prompts:
"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:
"Which are relevant? Act or watch?"
Agent compiles all selections into a risk register:
"Here's your de-risking plan for [product idea]:
| # | Dimension | Risk / Test | Recommended First Step |
|---|---|---|---|
| 1 | [e.g., Desirability] | [e.g., Gauge initial interest via landing page] | [e.g., Launch a 1-week landing page test with email capture] |
| 2 | [e.g., Legal] | [e.g., Data privacy compliance for EU customers] | [e.g., Map GDPR requirements to your data architecture this week] |
| ... |
| # | Dimension | Risk / Signal | What to Watch | Review Cadence |
|---|---|---|---|---|
| 1 | [e.g., Economic] | [e.g., VC funding contraction in AI] | [e.g., Quarterly VC reports, competitor funding rounds] | Quarterly |
| 2 | [e.g., Social] | [e.g., Shifting sentiment toward AI in healthcare] | [e.g., Industry surveys, media sentiment, customer interviews] | Monthly |
| ... |
Agent then offers:
"What would you like to do next?
pestel-analysis skill for a deep external-forces divepol-probe-advisor to choose the cheapest test for your top riskContext:
Internal Risk Scan Results:
Product Outcome Metrics:
Business Outcome Metrics:
External Risk Scan Results:
Risk Register Output:
Act on immediately (5 items):
Start tracking (5 items):
Biggest gap: Legal/compliance — HIPAA review must happen before any prototype touches real patient data.
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.
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.
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.
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.