Rumelt's strategy kernel: how to write a good strategy and spot a bad one

Who should use it, and when not to

Use it when a plan is a page of goals and initiatives and nobody can say what problem it solves. Use it to review a strategy deck before a board meeting, or to start a strategy for a company, a product, or a single team. It is also a good step before setting OKRs, so the objectives follow from a strategy instead of standing in for one.

It is less useful when the problem is already clear and the open question is which markets to serve and how to beat rivals there. That is the job of the strategy choice cascade. And if you first need to decide how to plan in a fast-changing market, see strategy styles for your environment.

The kernel in plain words

Diagnosis. A strategy is a response to a difficulty. The diagnosis names that difficulty and says why it is hard. It leaves out the side issues, so a reader finishes it knowing which part of the problem to push on. "Sales are down" is a symptom. "New customers leave before the habit forms" is a diagnosis.

Guiding policy. The overall approach chosen to deal with the obstacle in the diagnosis. It is a method, not a target. It directs effort and rules out other paths, without listing every step. If a policy rules nothing out, it isn't doing its job.

Coherent actions. Concrete steps, with people and money behind them, that are coordinated so each one makes the others work better. Rumelt illustrates this with Nvidia's early years: a diagnosis about the pace of the graphics chip market, a policy of releasing new chips faster than the industry norm, and actions such as overlapping development teams and heavy investment in simulation that made the fast cycle possible.

The four signs of bad strategy

A practical fix for the last sign: set one near-term objective the organization can reach with what it has, and that moves it through the obstacle. Ambition stays in the aspiration; the objective is the next step you can actually take.

A worked example

An illustration with a made-up company. A meal-kit service in Osaka has 2,400 subscribers. Last year's strategy read: "Become Kansai's most customer-centric food platform," with six goals: 30% subscriber growth, a corporate lunch line, vegan boxes, a Kobe launch, an app, and better NPS. Subscribers grew 4%. The data: 55% of new subscribers cancel within 8 weeks; those who stay past 8 weeks stay about 14 months. Exit surveys say cooking takes longer than the promised 30 minutes.

• Diagnosis: growth is stuck because new subscribers leave before the habit forms, and the cause is cooking time. The problem is the first 8 weeks, not reach.
• Guiding policy: make the first 8 weeks reliably easy. No Kobe launch, no lunch line, no app this year.
• Coherent actions: rebuild the starter menu around recipes tested under 30 minutes; pre-cut the slowest ingredients in starter boxes; contact every new subscriber in weeks 2 and 5; report cancellations by week and reason every Monday.
• Near-term objective: lower 8-week cancellations for next quarter's sign-ups to a target the team believes it can hit.
• Bad-strategy test on the old plan: fails all four. The slogan is fluff, the drop-off is never named, the 30% target has no "how", and six unranked goals pull the 18-person team in six directions.

How to write a strategy kernel

1. Gather the facts and the current plan. Write the situation in a few lines with numbers and their sources, and copy the existing goals as they are written.
2. Write the diagnosis. Name the one or two aspects that explain the situation and why they are hard. If two diagnoses compete, write both and say what evidence would settle it.
3. Write the guiding policy. One or two sentences on the approach, plus a line on what it rules out.
4. Choose three to five coherent actions. Each with an owner and resources. Say how each supports the others, and drop the ones that don't connect.
5. Set one near-term objective. Reachable with current resources, with an owner and a date.
6. Run the bad-strategy test. Mark each of the four signs Pass or Fail, quote the failing line, and rewrite it.
7. End with the decision. What the decider must approve, what stops because of the policy, and which facts are still to check.

The strategy kernel skill follows these steps and returns the kernel and the test on one page.

Common mistakes

• Starting from goals. Begin with the obstacle; goals come after.
• A diagnosis that lists everything. A list of ten problems is not a judgment. Pick the one that explains the rest.
• A policy that rules nothing out. If every current project still fits, the policy is too broad.
• Keeping every initiative. Coherence costs good ideas that pull in other directions.
• Vision-statement language. If a line would fit any company, cut it.

Where it comes from

Richard P. Rumelt, a professor at UCLA's Anderson School of Management, set out the hallmarks of bad strategy and the kernel of good strategy in "The perils of bad strategy" in McKinsey Quarterly in June 2011. His book Good Strategy Bad Strategy: The Difference and Why It Matters was published by Crown in July 2011. This page is an independent summary in our own words; it is not endorsed by the author or the publisher.

Questions

What is the kernel of a good strategy?

Rumelt's three parts, in our words: a diagnosis (what the real problem is and why it is hard), a guiding policy (the one way you will attack it), and coherent actions (a few steps that back each other up).

What are the four hallmarks of bad strategy?

Fluff, failure to face the challenge, mistaking goals for strategy, and bad strategic objectives. Rumelt set them out in The perils of bad strategy (McKinsey Quarterly, 2011) and in Good Strategy Bad Strategy.

Is a goal like 20% growth a strategy?

No. A target says where you want to be, not how you will get past what stops you. Rumelt counts treating goals as strategy as one of the hallmarks of bad strategy.