What Strategy Is
This is a typological theme organising three competing accounts of what strategy actually is — synthesised from Hamilton Helmer on 7 Powers, Roger Martin on Strategy, Richard Rumelt on Good Strategy Bad Strategy, Alex Komoroske on Strategy and Complexity, Jeetu Patel on Cisco, AI at Enterprise Scale, and the Right to Win, Gibson Biddle on Product Strategy, Ravi Mehta on the Product Strategy Stack, Twelve PM Competencies, and Selective Micromanagement, Seth Godin on This Is Strategy, the Four Critical Choices, and Why a Brand Is a Promise, Will Larson on Engineering Strategy, Systems Thinking, and Writing Consistently, Christopher Lochhead on Category Design, the Better Trap, and Languaging, and Melissa Perri on Strategy Deployment, the Missing Middle, and When to Hire a CPO.
The organising axis
Three accounts of what strategy is recur across the wiki. They are not incompatible at the surface — all three agree on the same negative case — but they diverge sharply on what the positive work of strategy involves and when advantage can be secured.
Where all three agree. Strategy is not goal-setting. A list of ambitions, profit targets, or market-share aspirations is not a strategy — it is the absence of one. Richard Rumelt calls the substitution of goals for diagnosis ‘bad strategy’ and traces most corporate strategy failures to it. Roger Martin makes the same point: ‘play to win’ means different choices than ‘play to play,’ and most companies never make that distinction. Seth Godin reframes it at the product level: founders treat the four choices that determine a product’s entire future — customer, competition, validation source, distribution — as constraints rather than decisions. The shared claim is that strategy is hard choice plus deliberate focus, not aspiration.
Where they diverge. The three accounts disagree on whether structural advantage can be designed in advance or only discovered by moving, and on whether the primary work of strategy is diagnosis, structural positioning, or disciplined navigation of an unpredictable system.
Account 1 — Structural power: advantage is designed
The first account holds that the purpose of strategy is to secure a durable structural advantage — a position competitors cannot replicate without self-harm. Advantage is not stumbled upon; it is designed, and there is a finite taxonomy of the forms it can take.
Hamilton Helmer’s 7 Powers is the clearest statement of this view. Power requires two conditions simultaneously: a benefit (above-average returns from a structural attribute) and a barrier (a reason competitors cannot replicate it). Operational excellence is not power — it is a treadmill every competitor must run and which produces no durable margin. The seven forms of genuine power — counter-positioning, scale economies, switching costs, network economies, branding, process power, resource power — each has a specific mechanism, and each becomes available at a different stage of company life. Counter-positioning is the only power accessible to a startup at inception; branding and process power require years. Founders who defer the question to post-PMF lose the counter-positioning window.
Jeetu Patel’s Right to Win asks the same question from a different angle. Permission to play — having the technology, people, and capital to compete — is necessary but not sufficient. The right to win is an unfair advantage that makes it logical that you specifically are the builder. Cisco’s right to win in AI networking derives from owning the infrastructure layer that has moved the world’s packets for forty years. A company without that structural anchor dissipates product spend regardless of quality. In an era when code is cheap to produce, the scarcest inputs are judgement about which problems to solve and a defensible reason you will win them.
Gibson Biddle’s DHM Model applies the same logic at the product level: every product strategy must delight customers in hard-to-copy, margin-enhancing ways. The hard-to-copy dimension — proprietary data, network effects, brand trust, switching costs, economies of scale — is the product-level equivalent of Helmer’s power taxonomy. Strategies that are delightful but easy to copy produce short-term gains competitors erase in months.
Roger Martin’s Strategy Choice Cascade gives this account an execution structure: winning aspiration → where to play → how to win → capabilities → management systems. The fifth layer — management systems that build and maintain capabilities over time — is what makes the moat real. Southwest Airlines’ single-aircraft model, flexible unionised labour, and point-to-point routing have not been replicated by any major US carrier in fifty years, not because no one has tried but because the capability stack is multifaceted enough to be uneconomic to replicate. The more interlocking the capability set, the less likely direct imitation becomes.
Account 2 — Deliberate choice: strategy is diagnosis
The second account holds that the primary work of strategy is accurate diagnosis, not positional building. Advantage follows from seeing the situation more clearly than competitors and making coherent choices that others avoid.
Richard Rumelt’s kernel — diagnosis, guiding policy, coherent actions — names what is missing from most strategy documents. A real diagnosis names why the problem exists, not merely that a gap exists. Without it, any guiding policy is arbitrary, and coherent actions are impossible to derive. The crux of a strategy challenge — the hardest move, whose resolution unlocks everything else — is found by immersion in the problem, not by selecting a framework from a list. Nokia’s failure to build a smartphone was not a resource failure; it was a diagnostic one. The CEO knew what was needed; the organisational structure made the diagnosis politically unusable.
Seth Godin’s Four Critical Choices make the diagnosis concrete at the product level. Choose your customers — the smallest viable audience, specified precisely — and you have implicitly decided the product, the future, and the competition. Trying to serve everyone is not a strategy; it is the refusal to make one. The four choices are interdependent: change distribution and you change the audience reachable; change the audience and competition looks different; change the validation source and what counts as success shifts entirely.
Ravi Mehta’s Product Strategy Stack applies the same logic to the planning hierarchy. Mission → company strategy → product strategy → roadmap → goals. Goals belong at the bottom, derived from the roadmap, not set at the top as OKRs from which work is reverse-engineered. Goals set without a roadmap are guesses; goals set after are commitments that honest people can be held to. The stack makes the diagnostic sequence explicit: you must know what you are going to build before you can say what winning looks like.
Melissa Perri identifies the most common failure mode not in strategy formulation but in translation. Executives have a strategy; teams have tickets; nobody is converting intent into team-level action. She calls this the Missing Middle. The diagnostic question — ‘What is the most important thing you could be doing, and why?’ — exposes the gap faster than any process review. When answers do not ladder to company strategy, the problem is not execution; it is deployment.
Christopher Lochhead’s Category Design adds a counterintuitive diagnostic: competing in an existing category is itself a strategic error. In venture-backed technology, one company earns approximately 76% of total category market capitalisation. Fighting for ‘better’ means competing for the remaining 24%. The diagnostic move is to restate the problem in a way that makes the existing solution inadequate — to frame, name, and claim a new category rather than accept the terms set by the incumbent. This is diagnosis as market design.
Account 3 — Emergent navigation: advantage is discovered
The third account holds that in complex, unpredictable systems, strategy cannot be designed or diagnosed in advance. The value is not in the plan but in the discipline of incremental movement: take the smallest defensible step, observe the reconfiguration, take the next step.
Alex Komoroske’s account of strategy in complex organisations is the clearest statement of this view. The Adjacent Possible is always smaller than assumed — the set of actions genuinely reachable from a current position. Each action reconfigures the world and reveals the next set, which will be different from the one before. Attempts to jump past the adjacent possible into long-range commitments are the primary source of strategic waste. This is not timidity but sequencing: slice decisions into the smallest defensible steps, each of which pays for itself. The failure mode is false precision — spending significant resource computing five-year numbers to two decimal places on decisions where the real question is order-of-magnitude.
But adjacent-possible alone produces pure incrementalism — gradient descent into a local maximum. Komoroske pairs it with a North Star: a low-resolution 3–5 year direction that every relevant domain expert finds plausible and worth high-fiving if achieved. The combined model is neither pure opportunism nor waterfall planning: at each decision point, find the adjacent action with the steepest gradient toward the North Star and take it.
Will Larson’s Engineering Strategy converges on the same insight from the engineering organisation. Every company already has an engineering strategy — it is visible in the consistent choices made about tools, architecture, and hiring. The complaint ‘we have no engineering strategy’ is always wrong; the strategy is just unwritten and therefore unimprovable. Writing it down makes it legible enough to debug. Good engineering strategies are often deliberately boring — Carta’s standard-kit constraint, Uber’s no-cloud policy, Stripe’s Ruby monolith — because boredom signals that engineers are not distracted by infrastructure choices and energy is focused on the problems the company cares about. Larson applies Rumelt’s kernel directly: a good engineering strategy starts with an honest diagnosis of the current situation, including its real constraints, which bad strategies habitually refuse to accept.
The emergent account does not reject structural advantage. It holds that advantage is discovered through movement, not secured in advance through design. The gardener does not plan the full tree from the acorn; they plant the seed, respond to growth signals, and look back at the compounding path after the fact.
Where the accounts genuinely conflict
The sharpest disagreement concerns timing and design.
Helmer holds that founders who defer strategic thinking lose the counter-positioning window — the only power available from inception. Strategy is something to think about from day one, tilting the odds before they are needed. The implication is that structural positions can be designed in advance and that deferring design is a cost.
Komoroske holds that the adjacent possible is too small for long-horizon structural design to be meaningful at early stages. The value is in the quality of the next step, not in a positional blueprint built before the terrain is visible. The gardener’s strategy ‘looks like luck’; it is actually farming for miracles. What Helmer calls ‘counter-positioning’ Komoroske might describe as the output of a successful series of adjacent steps that happened to put the company in an asymmetric position — recognised and named retrospectively.
Rumelt’s position sits between them. The kernel requires honest diagnosis, which is impossible without some immersion in the problem. But diagnosis precedes guiding policy, which in turn precedes coherent action — there is no such thing as purely emergent action without some framework for what the actions are for. The inert strategy — where guiding policies exist but no one implements them — is Rumelt’s specific failure mode, and it is exactly what Komoroske’s adjacent-possible discipline is designed to prevent.
The resolution the wiki’s sources do not fully supply: whether the three accounts are genuinely competing or are stage-dependent — emergent navigation appropriate early, structural positioning once a market position is legible, and deliberate-choice diagnosis a constant throughout.