Imagine I tell you our product strategy is: Increase retention. Expand enterprise revenue. Improve margin. Grow internationally. Use AI to increase productivity. All reasonable. I still do not know what you think is happening. I do not know what is blocking retention. I do not know where the enterprise opportunity actually lives. I do not know why margin is weak. I do not know what advantage we expect to create with AI.
I know what we want. I do not know the strategy.
Goals describe the destination
Goals matter. OKRs matter. Priorities matter. A roadmap can matter. The mistake is asking those artifacts to do a job they were not designed to do. “Increase retention by 10 percent” tells the organization what success looks like. It does not tell the organization why retention is weak or why one approach should work better than the alternatives. What is still missing is the explanation of why this approach should work better than the alternatives. That is where strategy begins. Richard Rumelt’s strategy kernel is useful because it forces three pieces of reasoning into the open:
What is the challenge? What approach are we choosing to address it? What coordinated actions make that choice real? The labels are less important than the logic.
Diagnosis is where strategy gets real
A real product environment is messy. Churn is up. Support volume is high. Customers complain about onboarding. The product has reliability problems. A competitor lowered its price. Sales is discounting more heavily. Usage is uneven across segments. You could respond to all of it. Strategy starts by simplifying. Which part of the mess explains enough of the problem that concentrating there should create disproportionate progress? That is diagnosis. Not because the other facts become false.
Because resources are finite. A useful diagnosis makes some work less important. If everything remains equally important after the strategy conversation, the strategy did not simplify the problem enough.
The guiding choice should change behavior
Suppose the goal is to improve on shelf availability in a large retail environment. That is a meaningful outcome. It is not yet a strategy. The problem could be inaccurate inventory data. It could be poor task prioritization. It could be labor availability. It could be product location accuracy. It could be execution quality in the aisle. It could be all of them. The strategic work is deciding what system of constraints matters most and where leverage sits.
If the diagnosis says this is a coordinated data, labor, prioritization, physical store, and product system problem, then “build another app” is a weak policy. A portfolio approach that connects those mechanisms becomes more coherent. Now the strategy changes what gets funded, what gets connected, and what teams should stop treating as isolated feature work. That is the test. Knowing the strategy should change behavior.
Coherence is evidence
This is the part I look for after somebody presents a strategy. Do the initiatives reinforce one another? Or are they just several good things grouped under the same objective? A retention strategy might include improving activation, fixing a reliability issue that disproportionately affects new customers, changing onboarding support, and removing a pricing surprise that creates early distrust. Those actions can form a coherent response to one diagnosed problem. Or the plan might contain a loyalty program, a redesign, three AI features, international expansion, and a new dashboard because those are the projects already in flight.
Same goal slide. Different strategy quality. Coherent action is where the theory becomes visible.
A strategy should be able to be wrong
This is one of the things I like most about explicit strategy. It creates something you can learn from. If the diagnosis is wrong, evidence can disprove it. If the guiding approach does not create the expected effect, you can revise it. If the actions do not reinforce one another, you can see the fragmentation. Vague aspiration is harder to falsify. “We need to delight customers” survives almost any result.
“Our retention problem is primarily caused by customers failing to reach first value during the first two weeks, so we will concentrate on activation and early workflow success before adding new retention programs” can be wrong. Good. Now there is a theory to test.
Do not turn the kernel into another template
There is an obvious failure mode here. The organization learns three strategy words and fills in three boxes. Diagnosis: “Customers want better experiences.” Guiding policy: “Be customer obsessed.” Coherent actions: every roadmap item. Nothing changed except the headings. The purpose is not vocabulary compliance. The purpose is forcing the causal logic into the open. What challenge are we really concentrating on? Where do we believe leverage sits? What are we doing because of that belief?
What are we not doing?
Goals are not the enemy
You still need goals. A strategy without an outcome can become elegant analysis with no accountability. But a goal without strategy leaves teams to invent the causal logic independently. That is how five teams can all pursue “retention” through five unrelated theories of the problem. So take the sentence your organization calls product strategy and ask one question:
What challenge does this diagnose?
If you cannot answer without adding information the strategy never states, you may have a list of things you want. That is useful. It is not yet a strategy.