Full SKILL.md

Strategic Choice

Frames a real either-or, generates three to five distinct strategic options, writes what would have to be true for each with the sceptic, tests the shakiest conditions first, and records the choice as a Playing to Win cascade with a Porter activity-system map and trade-offs. Use whenever a client must choose between enterprise-level alternatives it cannot all pursue, or a strategy excludes nothing, or the user says "which option should we pick", "three paths, can't decide", "what's our right to win", "is this strategy coherent", "what would have to be true", "we're everything to everyone", "specialise or stay broad", "write the strategy cascade". Product-specific where-to-play and proposition choices go to product-strategy, which may call this skill for the option test.

Replaces pros-and-cons debate with conditions that could be false, tested in order of least confidence. The Playing to Win cascade, Porter's positioning bases and fit tests, and the "lazy man's" sequencing are in references/frameworks.md; read it only when a step needs defending.

Routing

Not for: an undefined problem (problem-framing first); a missing fact base (market-competitive-intelligence or boardroom-storm-research first); a chosen strategy needing rival response (war-gaming) or execution (execution-planning).

Inputs

Ask for, or infer and label: who decides, by when, what triggered it; the fact base (segment size and profitability, customer research, relative cost and capability, competitor profiles); the strategy as practised, read from where revenue and profit come from by segment, channel and line (build a rough segment P&L with value-driver-modeling if none exists; options without it are stories without a baseline); constraints that will not relax; options already on the table, including the one someone senior is attached to; what was tried and abandoned.

Ask one round of questions if the decision-maker or the segment economics are missing. Otherwise proceed on labelled assumptions; Stage 3 will show which missing facts matter.

Workflow

  1. Frame a real either-or. One question with two or more answers that cannot both be taken; the status quo stated as a deliberate option with its own trajectory. Operational-effectiveness programmes (cost-out, digital parity, service levels) are worth doing under any option, so they are not options; if that is all that is on the table, widen the frame. Done when the decision-maker agrees picking one forecloses the others. Test: if we picked A, what could we no longer do?

  2. Generate distinct possibilities. Three to five options as three-sentence happy stories (where it plays, how it wins, why a customer chooses it), each with an advocate and an explicit "we would stop doing" line with its current revenue or attention cost. Generators: variety-based (which subset could we deliver better than anyone), needs-based (which group's needs justify a tailored activity set), access-based (who is reachable in a way rivals find uneconomic), status quo executed well; at most one hybrid with its own logic. No evaluation yet. Test: could a competitor pursue two at once without straddling? Then they are not distinct.

  3. What would have to be true (references/templates.md §3). Per option, six to twelve affirmative, falsifiable conditions under four headings: industry (size, growth, structural returns); customer value (they would switch and pay); relative position (cost and capability beat rivals, or can be built in time); competitor reaction (rivals will not or cannot remove the advantage). Ask the most opposed person "what would have to be true for you to support this?" and record the answer verbatim as a condition. Rewrite vague conditions until a specific fact would kill them; delete conditions that cannot be false; rewrite initiatives as the facts they depend on. Done when the sceptic confirms their objection is on the list. Test: which condition, if false, kills the option outright?

  4. Barrier conditions. Each team member rates confidence in each condition privately (high/medium/low); disagreement is signal. The barrier is the lowest-confidence condition that would kill the option. Record who is least confident; they design the test. Test: if this condition were true, would the sceptic support the option?

  5. Tests the sceptic will accept (references/templates.md §4). Per barrier: test (interviews, pilot, cost model, market data), pass threshold agreed before the data, sample or scope, owner, due date, time and cost. Cheap and fast on the least-confident condition beats comprehensive analysis. Competitor-reaction conditions go to war-gaming; driver-model conditions to value-driver-modeling. Test: pass switches the sceptic; fail drops the advocate. Both must be yes.

  6. Run tests, least-confident first. Stop testing an option once a barrier fails. Record results as data against the threshold, not interpretation. Untestable in time: carry as an explicit assumption with a signpost. Test: any threshold moved after the data arrived contaminates the result; say so.

  7. Choice and cascade. Five answers, each excluding something: winning aspiration (measurable, in customer and competitor terms); where to play (customers, geographies, categories, channels, vertical stages, and what is out); how to win (cost or differentiation, positioning base, why the chosen customer picks us over the best alternative); capabilities (three to six, existing vs to build); management systems (measures, processes, reviews that keep capabilities in place and signal when a condition stops holding). Check coherence bottom-up. Test: could a competitor say the same where-to-play and how-to-win sentences?

  8. Activity system and fit. Three to five strategic themes; concrete activities under each (done in a particular way, not departments); links between activities. Fit tests: consistency (no activity contradicts the position); reinforcement (three named pairs with mechanism); optimisation (three activities stopped, cost saved); interlock (the hardest-to-copy activity and the two that protect it; "our people" fails); deletion (remove a theme and the rest lose their logic). Trade-off register: what we will not do, why doing it breaks the system, what it costs now. Signposts: the conditions that, if they stop holding, reopen the choice, attached to management systems. Test: name the interlock or the advantage is not defensible.

  9. Choice record (references/templates.md §6). Framed choice; options and why each was rejected, citing the failed or unconfirmed condition; cascade; activity system; trade-offs; untested assumptions with signposts; decisions implied for the next 90 days. Done when a new executive could reconstruct the rejections from the record alone.

Output

The chosen option and its decisive condition in the first 150 words. Then, each once, 1,500 to 3,500 words:

  1. Choice record
  2. Framed-choice statement and option story cards
  3. Conditions sheet per option with barrier conditions marked
  4. Test plan and results log
  5. Cascade (five answers with exclusions)
  6. Activity-system map (text form) with fit-test results
  7. Trade-off register and signpost register
  8. Rejected-options table

No storyline (that is executive-communication), no competitor scenarios (war-gaming), no programme plan (execution-planning). No method vocabulary, framework authors, skill names or the worked example's client in the deliverable. Full example: references/worked-example.md.

Quality bar

Run silently; fix, do not report.

  1. The framed choice forecloses something.
  2. Status quo is an option with a trajectory, not a default.
  3. Every option has a stop-doing line and an advocate; none is a straw man.
  4. Every condition is affirmative and falsifiable; the sceptic's objection is on the list.
  5. Every barrier has a pre-agreed threshold and a sceptic-designed test.
  6. No threshold moved after the data.
  7. Each cascade answer excludes something a competitor could not equally say.
  8. Reinforcement pairs, stopped activities and the interlock are named.
  9. Every rejected option cites a condition, not an opinion.
  10. Untested assumptions carry signposts.
  11. No leaked method vocabulary.

Anti-patterns

  • Pros and cons resolved by seniority.
  • Operational-effectiveness programmes presented as strategic options.
  • Options that are compromises rather than positions.
  • Conditions written as questions, opinions or initiatives.
  • Advocate designs the test; threshold set after the result.
  • Continuing to gather evidence for a dead option.
  • Cascade answers a competitor could sign.
  • A strategy that adds activities and removes none.

Hand-offs

References

  1. Frameworks: Playing to Win and Porter's Strategic Coherence

    5,271 w
    references/frameworks.md9 sub-sections
  2. Templates: Strategic Choice

    1,534 w
    references/templates.md11 sub-sections
  3. Worked Example: Sablecrest Software

    5,117 w
    references/worked-example.md8 sub-sections