A goal is not a strategy. It is the thing a strategy is supposed to get you to. Those two keep getting swapped, so often that Richard Rumelt, the UCLA professor whose 2011 book Good Strategy/Bad Strategy is still the clearest writing on the subject, built an entire diagnosis around the swap. He tells the story of a graphics-chip company whose chief executive described the company’s strategy as “20/20”: twenty percent revenue growth a year, twenty percent profit margin. Rumelt’s verdict, which he laid out in McKinsey Quarterly, was that this was not a strategy at all. It was a statement of desire. It named a destination and said nothing about the terrain, the competitors standing on it, or how the company would actually cross it.
Open most product strategy decks and you find the same thing wearing a product costume. A revenue target. A north-star number. A list of initiatives for the next three quarters. Every item is real work and some of it is good work, but the document skips the one part that would make it a strategy rather than a budget with ambitions attached.
The four tells of a bad product strategy
Rumelt’s lasting contribution was naming what bad strategy looks like from the outside, so you can catch it in your own documents. He points to four hallmarks, and all four show up in product work.
The first is fluff. Language that sounds like insight and carries none: “deliver a seamless, delightful experience that puts the customer at the center.” Strip the adjectives and nothing is left to disagree with, which is the tell. A real strategic statement has an opposite that a reasonable person might have chosen instead.
The second is a failure to face the actual challenge. If the team is losing deals because onboarding takes three weeks and the competitor’s takes a day, the strategy has to say so, out loud, in the document. Most do not. Naming the problem means admitting you have one, and admitting it in a deck that goes to the executive team feels dangerous, so the hard sentence gets softened into a theme.
The third is mistaking goals for strategy. This is the 20/20 trap, and it is epidemic in product orgs that run on quarterly targets. A stack of OKRs is not a strategy, it is a scoreboard. I have written before about why most teams’ OKRs quietly stop working, and the root cause is almost always this: the numbers were set without a theory of what would move them.
The fourth is bad strategic objectives, which Rumelt describes as a long list of things to do, often labeled “priorities,” with no logic connecting them and no sense of which few actually matter. A strategy that will not say no to anything has not made a choice, and strategy is choice. The discipline of writing down what you have explicitly decided not to build does more strategic work than another quarter of initiatives.
The part that does the work is the diagnosis
Rumelt’s model of a real strategy has three parts he calls the kernel: a diagnosis, a guiding policy, and coherent action. The guiding policy is your overall approach. The coherent action is the set of coordinated moves that carry it out. But the part everyone skips, the part that makes the other two possible, is the first one.
A diagnosis, in his words, defines or explains the nature of the challenge. It takes a messy situation and names the one or two things that actually matter, the obstacle that, if you moved it, would change everything downstream. It is not the goal and it is not the plan. It is the honest account of why the goal is hard right now.
He uses the example of a chain of timber merchants whose written strategy was the usual: grow revenue 25 percent in three years, become the customer’s first choice. The rebuilt version threw all of that out and started with a diagnosis: customers were defecting because stock was inconsistent across branches, so the thing they needed most, reliable availability of core materials, was the thing the company was least able to promise. That was not a pricing problem or a marketing problem, which is where the generic goals had been pointing. Once the diagnosis was right, the actions wrote themselves.
Product teams skip the diagnosis for the same reason the timber merchants did. A diagnosis forecloses options. The moment you commit to “we are losing because our activation flow assumes a sophistication our new buyers don’t have,” you have ruled out the twelve other initiatives that were fun to talk about. A roadmap full of possibilities feels safer and more generous than a single named obstacle. It is also, in Rumelt’s precise sense, not a strategy.
A strategy nobody can name is not guiding anyone
There is a measurable cost to running on goals instead of a diagnosis, and it shows up as an organization that cannot repeat its own strategy back to you. Donald Sull and colleagues at MIT Sloan surveyed 4,012 managers across 124 companies and found that only 28 percent could list three of their company’s strategic priorities. Among the top team, just over half could. Among the leaders who reported directly to that top team, it dropped to 22 percent, and in one company a third of the leaders charged with executing the strategy could not name even one priority.
The usual read on that finding is a communication failure, as if the strategy were fine and just needed a better internal memo. I think the finding is mostly downstream of the diagnosis problem. A goal is hard to carry around in your head because it gives you nothing to decide with. “Grow 20 percent” does not tell a PM which of two features to cut when the sprint slips. A diagnosis does, because it is a lens: once you know the real obstacle, every tradeoff gets measured against whether it moves that obstacle. People remember a strategy they can actually use, and they use the ones that help them decide. If your strategy cannot be named by the people meant to execute it, the problem usually is not the wording. It is that there is no diagnosis underneath to name. That absence is exactly what a strategy alignment audit tends to surface.
The distribution client whose strategy was three numbers
A few years into my Ops Harmony work, I took a fractional COO engagement with a mid-sized distribution business. Their strategy document, which the owner was proud of, was three numbers and a list: grow fifteen percent, hit a target margin, launch two new product lines. Every quarterly meeting reviewed progress against the three numbers, and every quarter the numbers moved sideways while everyone worked hard.
It took about a month of looking at the actual account data to see what the three numbers were hiding. Two large accounts were carrying the whole business, and underneath them sat a long tail of small accounts that each lost money once you loaded in the cost to serve them. The sales compensation plan paid on new logos regardless of margin, so the team kept adding unprofitable small accounts and calling it growth. The real diagnosis, the sentence that was nowhere in the strategy document, was this: we are mistaking revenue from accounts that lose us money for growth, and our own comp plan is paying people to do it.
Nothing on the initiative list touched that. Once we wrote the diagnosis down, the moves were almost obvious and most of them were about stopping, not starting: re-tier the accounts, change the comp plan, raise prices or exit on the tail, protect and deepen the two anchors. The fifteen percent target did not change. What changed was that we finally had a theory of how to reach it, and the theory told everyone what to stop doing. That is the difference a diagnosis makes. It converts a wish into a set of decisions.
Start from the obstacle, not the ambition
The practical move is smaller than it sounds. Before the roadmap, before the OKRs, before the initiative list, write one paragraph that names the single biggest obstacle standing between you and the goal, in plain language, backed by something you can point to. Not “improve retention.” Something closer to “sophisticated buyers renew at 90 percent and the SMB segment we chased all year renews at 40, and our roadmap has been built almost entirely for the segment that is leaving.” If you cannot write that paragraph, you do not have a strategy yet. You have a goal and a hope, and no amount of initiative-planning will turn one into the other.
A roadmap that is really a feature list reads as generous because it promises everyone something. A strategy built on a diagnosis reads as narrow because it has decided what the fight actually is, and narrow is the point. The test of whether the pieces hang together, whether the actions actually follow from the obstacle you named, is the same one I described in the strategy narrative test: if you cannot tell the story from problem to action without a leap, the strategy is not coherent yet, and the order you tackle the work in is itself a strategic choice worth making deliberately.
Rumelt’s whole argument comes down to one uncomfortable idea: most of what gets called strategy is an attempt to skip the hardest thinking, which is figuring out what is actually in your way. The goals are easy. The list of initiatives is easy. Naming the obstacle, out loud, in a way that rules things out, is the work. Do that first, and the rest of the document starts to earn the word strategy. Skip it, and you have written 20/20 with better fonts.