Backstage · November 2026
32 Claude skills for product strategy
I’ve put together 32 Claude skills that take a product strategy from a fuzzy question all the way to a board memo. Each skill is one framework, written as instructions Claude follows: what to ask you, how to work, the exact output template and the checks it runs before handing anything over.
They’re made for heads of product, product managers and founders who own a strategy and want a strong first draft of every piece, without handing over the decisions.
Free to use. One folder per skill, plus the company context and a ready-to-upload zip of each.
Version 1.2, November 2026. I’m testing the skills with product people right now. Want to help? Try them and give feedback.
Who they’re for
Most useful for
- Heads of product and CPOs who shape, present and follow up the strategy
- Product managers and product owners who need a solid basis for priorities
- Founders and CEOs of smaller companies without a strategy team
Also useful for
- Consultants and interim leads who need to structure a strategy quickly in a new organisation
- Product designers and researchers working where discovery meets strategy
- Leadership teams and boards, who get clearer material with visible assumptions and named owners
Less useful if
- Your team doesn’t use Claude, although the approach carries over to other tools
- You want AI to make the decisions. These are built for the opposite
- You have no material at all. You’ll get a good structure, with a lot of assumptions
Four rules built into every skill
Sourced or flagged
Each skill tells Claude to give every figure a source or mark it as an assumption, and never to invent market numbers, competitors or customer quotes.
Claude drafts, a person decides
Every strategic choice gets a named decision owner. Claude recommends and lays out the trade-offs, and is told never to record a decision as made unless you say it was.
Gaps are written down
Each output ends with open questions, so thin evidence is visible instead of papered over.
Dates are absolute
“Q1 2027”, not “next quarter”.
Start here: your company context
The 32 skills are written for a generic company. The company-context skill tells them how your company works, so every output uses your language, metrics, finance rules and decision owners, without editing 32 files.
Fill in context.md yourself, or ask Claude to “set up our company context”. Every other skill reads it first. Anything you leave as Not set falls back to the skill’s default, clearly marked as an assumption. It also switches skills on, off or to a lighter version depending on your stage, from idea to mature. Keep the finished file in a Claude Project, so every conversation there uses it.
What it holds
- Company basics
- Language and vocabulary
- Stage and skills in use
- Metrics and data sources
- Finance rules
- Governance and decision rights
- Planning calendar
- Industry rules and buying process
- Strategy on record
- Formats and templates
What you say in the conversation always wins. Then the company context. Then the skill’s own defaults.
Six steps, from question to board memo
Use them in order for a full strategy cycle, or pick the one you need today.
Understand the problem, market and customer
You know which decision the strategy feeds, and what the market and customers say, before you choose.
| Skill | What you get | Use it when |
|---|---|---|
strategic-question-frame | What you getThe one decision the strategy must answer, its owner, deadline, the options on the table and what evidence would change the answer | Use it whenYou’re starting a strategy cycle, or the brief is “we need a strategy” |
product-health-read | What you getA traffic-light read across acquisition, activation, retention, revenue and referral, plus qualitative signals | Use it whenYou need an honest baseline before choosing where to focus |
issue-tree | What you getA MECE breakdown of the core problem into testable sub-questions | Use it whenThe problem is big and the team keeps arguing past each other |
competitive-analysis | What you getCompetitor profiles, a positioning map and where each one is weak, every claim sourced | Use it whenYou need to know where you can win, not just who exists |
needs-based-segmentation | What you getSegments defined by the job and the need, not demographics, with size and fit per segment | Use it whenYour “target customer” is everyone |
opportunity-solution-tree | What you getThe outcome, opportunities from research, candidate solutions and assumption tests (Teresa Torres’ structure) | Use it whenYou have discovery findings and need to connect them to an outcome |
market-sizing | What you getTAM, SAM and SOM worked out top-down and bottom-up, then reconciled | Use it whenSomeone asks “how big is this?” and wants a number you can defend |
lean-canvas | What you getAsh Maurya’s nine boxes on one page, with the riskiest assumptions flagged | Use it whenYou’re testing a new product or segment quickly |
Explore options and make choices
Each bet says what would prove it wrong, and the not-doing list is written down.
| Skill | What you get | Use it when |
|---|---|---|
product-vision-board | What you getVision, target group, needs, product and business goals on one board (Roman Pichler’s format) | Use it whenThe team can’t say where the product is heading in one breath |
strategic-options | What you getAt least three genuinely different options plus “do nothing”, each with its logic, cost and risk | Use it whenOnly one option is on the table, and it came from the loudest voice |
decision-matrix | What you getWeighted criteria agreed before scoring, the scores, and a check of whether the winner flips | Use it whenYou need to choose between options and show your working |
strategic-bets | What you getEach bet with its belief, evidence, investment, leading signal and “this is wrong if [measure] by [date]” | Use it whenYou’ve chosen a direction and need to commit resources to it |
not-doing-list | What you getWhat you’re saying no to, why, what it costs you, and when you’d look at it again | Use it whenEvery strategy, because the trade-offs are the strategy |
Challenge assumptions and test the thinking
Someone argues with the plan before a quarter is committed to it.
| Skill | What you get | Use it when |
|---|---|---|
hypothesis-board | What you getEach hypothesis with its test, the evidence needed, a kill threshold and its status | Use it whenYou want research aimed at the decision, not at everything |
scenario-planning | What you getThree to four plausible futures built on the two biggest uncertainties, and how the strategy holds up in each | Use it whenThe plan only works if the market behaves exactly as hoped |
red-team-review | What you getClaude argues as hard as it can against the plan, separating fatal flaws from fixable ones | Use it whenBefore you present, so the hard questions come from Claude first |
pre-mortem | What you get“It’s a year from now and this failed. Why?” Causes ranked, with early warnings and mitigations (Gary Klein’s method) | Use it whenThe team is confident and nobody has said what could kill it |
Build the business case and define success
Finance gets a return with its assumptions shown, and every bet gets a way to measure it.
| Skill | What you get | Use it when |
|---|---|---|
business-case | What you getAn assumptions register, base, downside and upside cases, costs, payback, ROI or NPV with formulas shown, and the 2–3 assumptions that move the answer most | Use it whenYou need budget or headcount approved |
north-star-metric | What you getCandidate metrics tested against customer value, link to revenue, how movable they are and gaming risk, with the input-metric tree and counter-metrics | Use it whenThe team optimises five metrics that pull in different directions |
strategy-kpis | What you getKPIs each tied to a bet, with baseline, target, owner, cadence and guardrail metrics | Use it whenYou need to know whether the strategy is working, not whether features shipped |
Turn strategy into a roadmap
The roadmap stops being a feature list the board reads as promises.
| Skill | What you get | Use it when |
|---|---|---|
impact-map | What you getGoal, actors, impacts and deliverables (Gojko Adzic’s method), so every deliverable traces back to a goal | Use it whenThe backlog is full of things nobody can connect to an outcome |
now-next-later-roadmap | What you getProblems and outcomes, not features; confidence drops from left to right; no dates on Next or Later; each item traced to a bet | Use it whenThe board treats roadmap dates as contracts |
quarterly-okrs | What you getOutcome key results (no tasks), 3–5 per objective, checked against the strategy KPIs | Use it whenPlanning the quarter so it actually moves the strategy |
Communicate, review and adapt
Teams can repeat the strategy, and every choice has a named owner and a date to look at it again.
| Skill | What you get | Use it when |
|---|---|---|
recommendation-memo | What you getOne recommendation, the alternatives considered, risks, the ask and the decision owner | Use it whenYou need a decision from leadership, not a discussion |
product-strategy-document | What you getThe full narrative: diagnosis, where to play, how to win, bets, not-doing, metrics and roadmap, plus a test teams apply to incoming requests | Use it whenYou need the source of truth everyone points to |
one-page-strategy | What you getThe whole strategy on one page, with a one-sentence version people can repeat | Use it whenMost people will never read the full document |
product-strategy-deck | What you getA 10–14 slide outline with an action title on every slide and speaker notes | Use it whenPresenting the strategy to leadership or the company |
roadmap-review-deck | What you getAn 8–12 slide outline reporting outcomes against bets, not shipped features | Use it whenThe quarterly roadmap review with stakeholders |
board-memo | What you getBottom line first, decisions made (with owner) kept separate from asks of the board, sourced numbers and risks | Use it whenBoard prep |
strategy-decision-record | What you getContext, options, decision, owner, date, consequences, review date and status, like an architecture decision record | Use it whenA strategic choice is made and you want the why kept |
quarterly-strategy-review | What you getKeep, kill or double down on each bet, judged against its own “this is wrong if” condition, with owner sign-off | Use it whenThe end of every quarter |
strategy-change-note | What you getWhat changed, the evidence, what it replaces, who decided, and what teams do differently from which date | Use it whenThe strategy shifts and teams need to hear it the same way |
How to use them
Install. Each skill is a folder with one SKILL.md file. Add the ones you want as custom skills in Claude. The setup steps are in Use skills in Claude and How to create custom skills. The folder name is the name Claude uses to pick each skill up.
Run a full cycle. Set up company-context first, then start with strategic-question-frame and work through the stages in order. Every skill says what it builds on and what it feeds into, so Claude can carry the work forward: the bets from step 2 become the KPIs in step 4 and the keep-or-kill calls in step 6.
Or pick one. Each skill works on its own. If an input is missing, Claude asks for the two or three that matter most and marks the rest as assumptions.
A few prompts to start with
- “Frame the strategic question for our 2027 planning. We’re deciding whether to move upmarket.”
- “Size the market for AI bookkeeping for Nordic SMBs, top-down and bottom-up.”
- “Red-team this strategy doc before Thursday’s leadership meeting.”
- “Build the business case for the self-serve tier. Finance wants payback and the assumptions.”
- “Turn our bets into a now-next-later roadmap the board won’t read as dates.”
- “Run the quarterly strategy review against our three bets.”
What stays with you: Claude drafts every piece. You, or the named owner, make every strategic choice.
