During my career I have read and been part of creating a lot of strategy documents. Different types of decks, one-pagers, so called "strategy houses," OKR cascades and vision papers. Some of them are sharp. Most of them are not. And the ones that aren't almost always fail in the same place, on the same missing pieces.
Richard Rumelt named those pieces better than anyone in Good Strategy / Bad Strategy. He calls them the kernel of a strategy: a diagnosis, a guiding policy, and a coherent set of actions. Three parts. Not optional. If even one is missing, you don't have a strategy. You have something that looks like one.
In my experience, most "strategies" in real organisations have zero of the three.
What the kernel actually is
Rumelt's kernel is short enough to fit on a napkin. Real strategies have:
A diagnosis. A clear-eyed read of the situation you're actually in. Not the situation you wish you were in. Not the slide that says "the market is changing." A specific account of what's happening, what's hard about it, and why the obvious responses don't work.
A guiding policy. A chosen approach for dealing with the diagnosed situation. Not a goal. An approach. It says how you're going to engage with the problem and, by implication, how you're not going to engage with it.
A coherent set of actions. Concrete moves that follow from the guiding policy and reinforce each other. Not a wish list. Not every project anyone wanted to do anyway. A set of actions that only make sense together.
That's it. Three parts, each one constrains the next. The diagnosis tells you what game you're playing. The guiding policy tells you how you're playing it. The actions are the moves in that game.
When you read it like that, it sounds obvious. So why do so few real strategies have any of these parts?
What most strategies have instead
Most strategies I see in practice have one thing: a list of goals, sometimes dressed up as "strategic priorities."
Grow revenue 20%. Improve customer satisfaction. Become more data-driven. Win in segment X. Launch the new platform. Reduce time to market. Build a customer-centric culture.
That's not a strategy. That's a list of things the leadership team would like to be true. There's no diagnosis ("why isn't revenue already growing 20%?"), no guiding policy ("what is our actual approach to making it happen?"), and no coherent actions ("which moves only make sense given that approach?").
Rumelt has a sharp word for this. He calls it bad strategy. The goals might be perfectly reasonable, but it's bad strategy because it skips the work. It hands the hard part, the choosing, to the people below, and expects them to figure it out.
Spoiler: they can't. And the reason is the choosing isn't theirs to make. Without a diagnosis and a guiding policy from above, every team picks its own interpretation of the goals, optimises for what its own incentives reward, and the organisation ends up running in slightly different directions while everyone reports green status.
Why the diagnosis is the part everyone skips
Of the three pieces, the diagnosis is the one that gets dropped first. And usually it's not even noticed.
Diagnosis is hard because it requires admitting something is wrong. Not in a vague "we need to evolve" way, but in a specific "here is the thing that is broken, and here is why we haven't been able to fix it" way. Oftern times leadership teams resist this. It feels like blaming. It also feels like risk: if you write down a clear diagnosis, you've just made yourself accountable for whether your read of the situation was right.
So instead of a diagnosis you get a description. "The market is becoming more competitive." "Customer expectations are rising." "Digital transformation is accelerating." None of those are diagnoses. They can be compared to weather reports. They tell you nothing about what your organisation specifically should do.
A real diagnosis sounds different. It sounds like: "We're losing mid-market customers because our onboarding takes six weeks while three competitors do it in three days, and the reason it takes six weeks is that the legal review of every contract still goes through one team in Helsinki." That's something you can build a guiding policy around. "The market is becoming more competitive" is not.
If you cannot write down the problem in a sentence that points to specific causes, you don't have a diagnosis yet. And without it, the guiding policy and the actions have nothing to hang on.
Why the guiding policy gets confused with vision
The second part, the guiding policy, gets confused with vision statements all the time. They are not the same thing.
A vision says where you want to go. A guiding policy says how you're going to get there given the diagnosis. Vision is "we will be the leading digital insurer in the Nordics." Guiding policy is "we will compete on speed of decision rather than breadth of products, by automating underwriting for the segments where claim risk is statistically predictable and exiting the segments where it's not."
You can hear the difference. The vision sounds inspirational. The guiding policy sounds like it costs something. That's because it does. It tells you what you're doing and, by direct implication, what you're not doing. A real guiding policy always closes doors. If your guiding policy doesn't make some part of the organisation uncomfortable, it probably isn't a guiding policy. It's just a slogan.
This is the same point Roger Martin makes about strategy as choices, and Michael Porter makes about strategy as trade-offs. Different vocabulary, same idea. If nothing has been chosen against, no real strategy has been made.
Why the actions have to be coherent, not just plentiful
The third part, the coherent set of actions, sounds like the easy one. It's not.
Most organisations are very good at producing actions. The annual planning cycle alone produces hundreds of them. The problem isn't quantity. It's coherence. A coherent set of actions means the moves reinforce each other. Each one only makes sense given the others. Pull one out and the others get harder. Add a random new initiative and the whole shape weakens.
In practice, what I see is the opposite. Strategy gets approved at the top, and then every function adds its own initiatives to the list. Marketing has its priorities, product has its roadmap, HR has its programmes and operations has its efficiency targets. The list grows. None of it cancels anything else out, because cancelling things is uncomfortable. By the time you get to execution, you have fifty initiatives, each defended by someone, none of them connected to the others by anything stronger than the fact that they are all on the same slide. Not to mention to that people responsible on executing these things have yet another items on their backlog todo and often times everything should happen at the same time.
A coherent set of actions is rarely longer than five to seven things. If your strategy has thirty initiatives, it doesn't have a kernel. It has a budget compromise.
The honest test
Here's the test I run when someone hands me a strategy document and asks me to help with execution. I ask three questions, in order.
One. What is the specific situation this strategy is responding to? Not the description. The diagnosis. What is broken, and why has it been hard to fix until now? If the answer is generic, the kernel is missing the first part.
Two. What is the chosen approach for dealing with that situation? Not the goals. The approach. What are we doing that's different from what we'd do if the diagnosis were different? If the answer is "we'll do all of these things," the kernel is missing the second part.
Three. Which of these initiatives only makes sense if we take that approach? If the answer is "all of them would make sense anyway," the kernel is missing the third part.
Most of the time, all three answers are missing. And once you see that, you understand why the strategy isn't landing in execution. There was nothing to land. The deck was approved. The work of strategy was never done.
What to do about it
If your current strategy has zero of the three, you don't need a new one. You need to finish the one you have.
Sit down with the leadership team and write the diagnosis. Don't do the polite version, but write down the version that names the actual thing that's blocking you. Then write the guiding policy that follows from that diagnosis. Notice what the guiding policy forces you to stop doing, and write that down too. Then look at your current initiative list and cut everything that doesn't reinforce the guiding policy.
You will end up with a much shorter list of actions and a much clearer answer to "what actually is our strategy." The list will be uncomfortable, because it will have removed things people like. That discomfort is the signal. It means the strategy now has a shape. Before, it had a wish list.
Rumelt's kernel isn't a framework you bolt onto a finished strategy. It's a test for whether you have a strategy at all. Most organisations fail it. The fix isn't new ideas. It's finishing the work that was started and then handed off too early.
I'm Jenni. Most strategies stay vague. I help organizations define what their strategy actually means in practice, then build the operating models, processes, and concepts to execute it. Founder of Digital Rebel.