Full SKILL.md

Strategy Toolkit


name: strategy-toolkit description: Use for general management-strategy questions that product-strategy does not cover: turnarounds, growth diagnosis, portfolio reallocation, competitive response, transformation, cost and margin, operating-model redesign, M&A rationale, or any ambiguous "what should we do" business question. Twenty-one short analytical moves for a single-session read, from situation assessment to decision memo, plus an 8-stage problem-solving workflow, in McKinsey, Bain and BCG idioms. Use whenever someone is diagnosing a business, generating or stress-testing strategic options, building the case, planning delivery, or packaging a recommendation, even if they do not say "strategy". For product, market-entry or monetization work use product-strategy; for a full-depth method with gates, templates and worked examples use the consulting-engine skills.

A reusable playbook for general management-strategy work: turning an ambiguous business question into a structured, defensible, board-ready recommendation. This file is the spine — it carries the partner-led workflow, the operating rules, and a map to the 21 analytical moves. Load a reference only when the work touches it.

This is the general sibling to product-strategy. Use product-strategy for product, market-entry, expansion, monetization, and GTM engagements (it has the where-to-play / what-to-build / how-to-win / how-to-make-money spine). Use strategy-toolkit for everything else a strategy team handles — turnaround, growth diagnosis, portfolio reallocation, competitive response, cost and margin, operating-model redesign, transformation, M&A rationale — and for the underlying analytical moves either engagement needs. The two share vocabulary (SCQA, MECE, the Pyramid, business cases, KPI trees) and cross-reference cleanly. Do not use for product or proposition questions (where to play, what to build, how it makes money, how to launch): that is product-strategy. Do not use when the work needs a gated method with templates and a worked example: that is the consulting-engine pack, and each move below names its counterpart under "Long form".

How to use this skill

  1. If you have an ambiguous question and a blank page, run the 8-stage workflow below — it sequences the whole engagement with review gates.
  2. If you need a specific analytical move (size a market, find the binding growth constraint, prioritize initiatives, build the case, align stakeholders), jump straight to it via the move catalog and load that one reference.
  3. The moves are written in common consulting vocabulary. To match a client's house idiom, references/firm-lenses.md re-flavors any move as McKinsey (issue trees, Pyramid, SCQA, Three Horizons), Bain (gap to full potential, Founder's Mentality, NPS/loyalty economics, RAPID, microbattles, Results Delivery), or BCG (Strategy Palette, Growth-Share, experience curve, TSR, DICE, Smart Simplicity).
  4. Hold the operating rules throughout, and run the relevant quality bar before treating a deliverable as done.

The 8-stage workflow (the engine)

A partner-led, hypothesis-driven problem-solving sequence. The user is the partner (sets direction, makes trade-offs, approves each stage); Claude works as the associate. Every stage has a review gate: present the stage output and get the user's approval before starting the next.

  1. Frame the question — restate the problem as a single decision question that, once answered, tells the client what to do. Apply SCQA; define scope and success criteria. Output: one-line decision question + SCQA + scope + success criteria.
  2. Map the logic — decompose the question into a MECE issue tree; attach a testable hypothesis to each branch; tag each known / assumed / unknown. Output: a hypothesis tree.
  3. Focus the work (80/20) — score each branch on how much it could move the answer × how uncertain it is; select the few high-impact, high-uncertainty branches; park the rest with a one-line reason. Output: a prioritized shortlist + a parked list.
  4. Plan the analyses — per selected branch, define the analysis, data, and method; sequence by dependency; flag data gaps; set the "done" bar. Output: an analysis plan.
  5. Run the analyses — test each hypothesis against evidence; record finding + evidence + confidence (high/med/low); revise where evidence contradicts; surface new branches. Output: findings with confidence.
  6. Integrate the answer — synthesize the branch findings into one governing answer; resolve tensions (don't average them away); pressure-test with "what would have to be true for this to be wrong." Output: one integrated answer with support and open risks.
  7. Shape the recommendation — convert the answer into concrete actions (what, by when, owned by whom), economics (value at stake vs cost), key risks + mitigations, and first moves. Output: a recommendation.
  8. Package the story — structure with the Pyramid Principle (governing thought → MECE supports → evidence), open with SCQA, produce in the requested format, prepare hostile-question answers (see narrative-builder). Output: board-ready content + a hostile-Q prep sheet.

Operating rules

  • Answer-first — lead with the recommendation, then the support. Suspense is a defect.
  • MECE — every decomposition mutually exclusive and collectively exhaustive.
  • Hypothesis-led — write the hypothesis before running the analysis.
  • 80/20 — spend effort only on the branches that can change the answer.
  • Evidence-graded — every claim carries a confidence level and its source; never smuggle an inference in as a fact.
  • So-what discipline — every finding ends in an explicit implication; if it has none, cut it.
  • Second-order thinking — for every major move ask "and then what happens" (competitor response, the next binding constraint, the strain on the core).

The 21 analytical moves — catalog and reference map

A typical engagement flows diagnosis → choice → execution → communication, but the moves are standalone — pull the one the moment demands. For deck/exhibit/board-page craft on the output, hand to the consulting-deck skill; for live .xlsx models behind a business case, the financial-modeling skill.