Ask most leadership teams for their strategy and you’ll get a list. Grow revenue 30%. Become the market leader. Win on AI. Delight customers.
Richard Rumelt would call that a set of aspirations, not a strategy. And the distinction is not academic. It’s the difference between organizations that move with force and organizations that spread themselves thin across everything at once.
Bad strategy has a shape, and it’s recognizable
In Good Strategy Bad Strategy, Rumelt does something rare. He doesn’t just describe what good looks like — he names what bad looks like, precisely enough that you start seeing it everywhere.
Bad strategy is fluff: abstractions dressed up as insight, the kind of sentence that sounds profound in a slide and means nothing when you try to act on it. It’s the failure to face the challenge — if you don’t name the actual obstacle, you can’t overcome it. It’s mistaking goals for strategy: stacking ambitious targets on top of each other as if wanting the outcome badly enough were a method for reaching it. And it’s bad strategic objectives — goals that don’t connect to any real problem, or that scatter effort across a dozen priorities that quietly cancel each other out.
The uncomfortable part: most strategy documents I’ve seen are some blend of these four. Not because the people writing them are weak, but because naming the real challenge is harder — and more exposing — than listing the outcomes everyone already wants.
The kernel: diagnosis, guiding policy, coherent action
Rumelt’s alternative is deceptively simple. Good strategy has a kernel — three parts, in order.
A diagnosis. What is actually going on here? Not the symptoms, not the wish list — the underlying structure of the situation. A good diagnosis simplifies a messy reality into something you can act on. It replaces “we need to grow faster” with “our growth is capped because onboarding takes six weeks and our best customers churn before they see value.”
A guiding policy. The overall approach chosen to deal with the diagnosis. Not a goal — a direction. It rules things out. A guiding policy is valuable precisely because it says we will do this and not that, which means it can be wrong, which means it’s real.
Coherent action. Coordinated steps that support the guiding policy and each other. Coherent means they point the same way. Most organizations fail here quietly — each initiative is defensible on its own, but together they pull in different directions and the force dissipates.
The power isn’t in any one element. It’s that they’re linked. A diagnosis with no policy is analysis. A policy with no coherent action is a slogan.
The crux: where importance meets addressability
Rumelt’s later work, The Crux, sharpens the point into a single word.
Strategy, he argues, is fundamentally problem-solving under conditions where you can’t do everything. And the skill that matters most is identifying the crux: the part of the challenge that is both important and addressable. Not the biggest problem — the biggest problem you can actually make progress on.
This is a discipline of subtraction. Any serious challenge has many hard parts. Some are important but immovable. Some are addressable but don’t matter much. The crux is the intersection — the point of maximum leverage, where concentrating your resources actually changes the outcome.
Great strategists don’t attack everything. They find the one knot that, once loosened, makes the rest of the tangle give way. Then they pour disproportionate energy into it.
Why this matters more in the AI era, not less
Right now, “win on AI” is doing exactly what Rumelt warned about. It’s a goal wearing the costume of a strategy.
Every organization has ambitious AI objectives. Very few have done the diagnostic work: what is the actual challenge AI creates or exposes for us specifically? Fewer still have a guiding policy that rules anything out. And almost none have found the crux — the one obstacle where concentrated effort would genuinely change their trajectory, as opposed to funding twelve pilots that each look reasonable and collectively go nowhere.
The temptation, under pressure, is to add priorities. Rumelt’s entire body of work argues the opposite: the leaders who win are the ones willing to say this is the thing that matters most, and here is what we are deliberately not doing about everything else.
Strategy is a choice, and choices exclude
The reason good strategy is rare isn’t that the framework is hard to understand. It’s that the framework forces you to choose — to name the real problem, to commit to one direction, to concentrate force on a single crux while leaving other worthy things unfunded.
That’s uncomfortable. A list of goals is safer. It offends no one, excludes nothing, and asks for no hard tradeoff.
But it also doesn’t work. And in a moment when everyone is accelerating in the same direction at once, the organizations that identify their crux — and act on it with coherence — will quietly pull away from the ones still polishing their list of ambitions.
This is the lens I keep coming back to in conversations with executive teams trying to turn AI ambition into actual strategy. If it resonates, I’d be glad to think it through with you. Find me on LinkedIn.