Rumelt's Good Strategy/Bad Strategy organizes every real strategy (a business one, a military one, a personal one) into what he calls a kernel. It has three parts, and all three must be present, or the thing calling itself a strategy isn't one yet.
What are the three parts of Rumelt's kernel?
Diagnosis simplifies the tangled, ambiguous, over-determined reality of a situation down to its critical, addressable structure. That doesn't mean everything true about the situation. It means the specific thing that, once named, tells you what actually has to be dealt with. Guiding policy is the overall approach chosen to cope with the obstacles the diagnosis identified. A goal or a hope doesn't count. Guiding policy is a stance that rules things out ("we will compete on X, which means we will not chase Y"), and it can do that because it follows from a specific diagnosis instead of an unconditional wish list. Coherent action is the set of steps that reinforce each other, are actually resourced, and are coordinated well enough to carry the guiding policy out. A slide of initiatives that could each exist independently of the others doesn't qualify.
Advertisement
How do frameworks fit with the kernel?
Now hold that structure up against the frameworks you've already learned. You came out of the last two framework sets with eleven named ones, and every one of them is, at most, a candidate tool for one part of the kernel. Usually that part is diagnosis. Five Forces organizes a structural read of an industry. 7S organizes an internal-alignment read, and jobs-to-be-done a demand-side read. Occasionally it's guiding policy, as when generic strategies propose a stance: cost, differentiation, or focus.
None of them is the kernel.
A team can run a textbook-perfect Five Forces, correctly conclude "supplier power is high and rivalry is intense," and still have produced zero diagnosis. "Supplier power is high" was true before anyone opened the laptop, and it stays true whatever the client's specific number is doing. It explains the industry. It says nothing about the client's own performance inside it. That's the failure you're here to learn to catch, and it has a name: framework-as-theater, a framework run competently, in the correct order, with the correct axes, that no actual decision ever rests on, because it was never testing anything that could have come out differently.
The four faces of bad strategy
Rumelt names four recurring shapes that bad strategy takes. Every one of them hides comfortably inside a well-executed framework, so "we ran the framework correctly" and "we have a strategy" are two different claims.
Fluff restates something simple in a fog of invented-sounding jargon and calls the fog insight. "We will harness our synergistic capabilities to unlock stakeholder value across the ecosystem" says nothing that would look different if it were false. A framework dresses fluff beautifully. Every quadrant, box, or force gets a plausible-sounding bullet, and the plausibility of the language gets mistaken for the presence of an idea.
Failure to face the challenge is a strategy that never actually names the real obstacle. Often that's because naming it assigns blame, admits a sunk cost was wasted, or threatens a business model everyone in the room is emotionally invested in. A framework makes this easy to hide. You can fill in "threat of substitutes: moderate" without ever writing the sentence "our core product is being substituted away, and here is the evidence."
Mistaking goals for strategy treats an ambition ("double revenue by 2029," "become the most trusted brand in the category") as if stating it were the same as having a plan for it. A guiding policy is a specific, falsifiable claim about how, given named obstacles. A goal is silent about both the obstacles and the how. And no framework's output supplies the missing "how" for you, however many boxes are filled.
Bad strategic objectives is a list that never says no to anything: eleven "strategic pillars" for the year, no ranking, no resource tradeoff named, nothing sacrificed. A plan that doesn't choose hands every department a permission slip to keep doing what it was already doing, and that's no strategy.
Every worked example in the full module shows at least one of these four faces wearing a framework's clothing. That pairing, a real bad-strategy pattern executing correctly inside a real framework's mechanics, is what "framework-as-theater" means in practice. It's also why this can't be taught as a bolt-on paragraph to the framework mechanics. Running the mechanics well is necessary, and you've already been taught it. It's nowhere near sufficient.