The observation
Ask most companies for their strategy and they hand over a document full of ambitions: grow revenue 20%, expand into three new markets, become the market leader, delight customers. It reads like strategy. It has the right vocabulary. And Richard Rumelt, who has spent a career diagnosing why strategies fail, calls this exactly what it is: bad strategy. Not a weak strategy, or a risky one — the absence of one, dressed up to look like the real thing.
Rumelt names four tells. Fluff — abstract language that sounds important and says nothing. Failure to diagnose — jumping straight from a problem statement to therefore, with no because in between. Goals masquerading as strategy — a profit target or a market-share number is an ambition, not a plan for getting there. And incoherent actions — a company told to grow faster and raise its return on equity at the same time, as if wanting both settles how to get either.
The counterintuitive core
The natural response to a hard business problem is to state how you'd like it to end. Rumelt's claim is that this feels like strategy and is not strategy — it is the refusal to do strategy's actual work. A real strategy, in his terms, has three parts, which he calls the kernel: a diagnosis of what is really going on and why, a guiding policy for dealing with it, and coherent actions that carry the policy out without contradicting one another. Strip out any one of the three and the rest is decoration.
Roger Martin, coming at strategy from a different angle — his five-question Strategy Choice Cascade — lands on the same distinction by another name. He separates 'play to win' from 'play to play': an aspiration built around beating specific competitors for specific customers, versus a company that shows up, competes, and hopes. Most companies, he argues, never notice they are doing the second while believing they are doing the first, because a goal and a genuine competitive aspiration read identically on a slide.
Seth Godin reframes the same failure at the level of a single product. He argues that four decisions — who your customers are, who your competition is, whose judgement you are trying to satisfy, and how you reach people — quietly determine everything else a product becomes. Most founders never choose any of them; they inherit all four and treat the result as a given. Not choosing is itself a choice, just an unexamined one — and it is why 'we serve everyone' functions as the goal-setting equivalent of no strategy at all.
The line the wiki's What Strategy Is theme draws across all three: strategy is hard choice plus deliberate focus, not aspiration. What differs between Rumelt, Martin and Godin is where they think the primary work lies — diagnosis, structural positioning, or product-level choice — not whether goal-setting alone qualifies as strategy. On that, all three agree: it doesn't.
What it means in practice
- Ask why, not just what. A diagnosis names the cause of the problem. 'Revenue is flat' is an observation. 'Revenue is flat because our distribution channel changed hands and we never rebuilt the relationship' is a diagnosis — the second one tells you what to do next; the first one doesn't.
- Find the crux, not the full list. Rumelt names the hardest sub-problem the crux, after the hardest move in a rock climb — the one whose resolution unlocks the rest. His consulting method surfaces 25–50 candidate problems on Post-its, then filters for which ones are both important and actually addressable, before picking one or two to act on.
- Call it an action agenda. Rumelt's own trick: strip the word 'strategy' from the document and ask instead, plainly, what are we actually going to do about this one challenge? The word 'strategy' invites fluff; 'action agenda' does not.
- Choose the four things you are pretending are fixed. Godin's version: name your smallest viable audience, your actual competition, whose approval you are chasing, and how people will find you — on purpose, before you build anything else. Each choice narrows and sharpens the other three.
- Decide whether you are playing to win. Martin's cascade opens with a winning aspiration framed around customer benefit, not market share, and a where-to-play choice — a specific battlefield, not every battlefield. Answering both honestly is what separates a plan for beating someone from a plan for showing up.
The honest limit
The three accounts agree that goals aren't strategy, but they don't converge on what to do instead. Rumelt's kernel makes diagnosis the primary act — get the 'why' right and the actions mostly follow. Martin's cascade makes structural position the primary act — capabilities and management systems that competitors cannot cheaply copy, built over years, the way no US carrier has replicated Southwest's model in fifty years. Godin's four choices sit somewhere between the two: less a search for a durable moat, more a discipline of not defaulting on decisions that are already yours to make.
None of the three tells you which framing to reach for on a given morning. A start-up choosing its first customers needs Godin's discipline more than Martin's capability stack, which takes years to matter. A hundred-year-old incumbent losing share needs Rumelt's diagnosis before either of the others is worth attempting. The wiki's sources leave that sequencing question open — useful to know going in, rather than discovering it only once the wrong framework has been applied to the wrong problem.
Go deeper
The single best source for this idea in full is Richard Rumelt on Lenny's Podcast — watch it here, or read the episode.