Good Strategy, Bad Strategy: Rumelt's Kernel Test as a Checklist

Action item 122: create a strategic plan

Richard Rumelt once sat down with the mayor of a small city in the Pacific Northwest to talk about strategy. The city's planning committee had produced a strategic plan. It contained 47 strategies and 178 action items. Action item number 122 was “create a strategic plan” ¹.

Nobody on that committee was lazy. They had collected every good idea in the room and written each one down. What they had not done was say what problem the city faced, which is the one thing that would have let them throw 170 of those items away.

The short answer. In Rumelt's framework, a good strategy has a three-part kernel: a diagnosis that explains the nature of the challenge, a guiding policy that sets an overall approach to overcoming it, and coherent actions that are coordinated to carry the policy out ¹. Bad strategy shows four hallmarks: failure to face the challenge, mistaking goals for strategy, bad strategic objectives and fluff ¹. His book, Good Strategy/Bad Strategy, develops both ideas at length ².

This page turns that framework into a checklist you can run on any strategy document. Its argument is simple. Bad strategy is rarely a wrong answer. It is an answer to no question: goals and action lists sitting where a diagnosis of the obstacle should be. So the checklist starts with the diagnosis, because every other item depends on it.

No obstacle, no strategy. Just a list of things you wish would happen.

The kernel, in working terms

Rumelt defines a strategy as a way through a difficulty, an approach to overcoming an obstacle, a response to a challenge ¹. The kernel is the minimum structure that makes that true.

Rumelt's worked example is Nvidia in the mid-1990s. Its first product, a PC add-in board, was a commercial failure. In 1995 rival start-up 3Dfx Interactive took the lead in fast 3-D graphics chips for gamers, and there were rumors that Intel would enter. The diagnosis: “We are losing the performance race.” The guiding policy, in Rumelt's summary: release a faster, better chip three times faster than the industry norm, every 6 months instead of every 18 ¹.

Then the coherent actions. Nvidia formed three development teams on overlapping schedules, invested heavily in simulation and emulation so that chip fabrication and software drivers would not cause delays, and over time took driver development back from the board makers ¹. Each action makes sense only because of the policy, and the policy only because of the diagnosis. Rumelt reports that Intel entered with its own chip in 1998 and left the discrete 3-D graphics business a year later, and that creditors of 3Dfx started bankruptcy proceedings in 2000 ¹.

He also adds a caution in a footnote: even a good strategy's effectiveness is not permanently assured ¹. The kernel tells you a document is a strategy. It does not tell you the strategy is right.

The checklist

Run these 14 checks on the document your company calls its strategy. For each one, find the evidence on the page and mark it pass, weak or fail. Do not give credit for things people say in the meeting. If it is not written down, it will not survive the first budget round.

Score it this way: any fail in checks 1 to 6 means there is no kernel yet, and the right next step is a diagnosis, not a better plan. Fails in 7 to 14 mean the kernel exists but the document has been diluted, and the fix is editing.

The four hallmarks, and a test for each

Rumelt condensed his list of bad-strategy signs to four ¹. Each has a quick test you can run in a meeting.

• Failure to face the challenge. In July 1979, International Harvester's planners produced a thick corporate strategic plan stitched together from five divisional plans. It projected agricultural-equipment share rising from 16% to 20% and a hockey-stick recovery in profit. It did not mention the company's grossly inefficient plants or its labor relations, which Rumelt calls the worst in US industry, even though, as a result, the company's profit margin had long been about half its competitors' ¹. Test: does the plan explain the number that is most embarrassing?
• Mistaking goals for strategy. A CEO Rumelt calls Chad Logan had a “20/20 plan”: 20% annual revenue growth and a profit margin of 20% or more. Asked what would have to happen to get there, he talked about the will to win ¹. Test: cover the targets. Is anything left?
• Bad strategic objectives. Two forms. The “dog's dinner” is a long list of things to do, swept in from a planning meeting and labeled long term so that none of it needs doing today. The “blue sky” objective restates the desired outcome and skips the fact that no one knows how to reach it ¹. Test: for each objective, ask whether the team could achieve it with the resources it has. If it is as hard as the original challenge, it adds nothing.
• Fluff. Rumelt quotes a major retail bank's internal strategy memo: “Our fundamental strategy is one of customer-centric intermediation.” Remove the buzzwords and the bank's strategy is being a bank ¹. Test: translate each sentence into words a new hire would use. If the translation is obvious, the sentence was fluff.

The goals hallmark gets a full treatment on the what strategy is not page, alongside four other things that get mistaken for strategy. This page stays with the kernel.

How to run the review

1. Get the real documentUse the version that went to the board or the budget process, not the offsite slides. The budget version is the one that decides where money goes.
2. Mark it up in three colorsHighlight every sentence that diagnoses (explains why things are as they are), every sentence that sets policy (chooses an approach) and every sentence that commits action (money, people, dates). Most documents come back almost entirely in the third color.
3. Run the 14 checksScore each pass, weak or fail, and write the evidence next to it: the sentence that passes, or the gap that fails.
4. Draft the missing diagnosis as a hypothesisIf checks 1 to 3 fail, write your best guess at the obstacle in two sentences, with the evidence for it. Make it specific enough to be wrong. Hypothesis-driven problem solving is the method.
5. Take the diagnosis to the owner, not the scorecardLead with the proposed diagnosis. People argue far more readily with a score than with evidence about the obstacle. The scorecard is your preparation, not your presentation.

A worked example: scoring a software company's strategy

Look at what the diagnosis did. Before it, “expand into enterprise” and “win in Europe” were as defensible as anything else on the list. After it, they are distractions from the leak. Nobody had to argue against them on their merits. The diagnosis made the argument.

Why smart teams write bad strategy

Rumelt names two roots: the inability to choose, and template-style planning ¹. The first is political, and the clearest example he gives is a meeting he attended at Digital Equipment Corporation in 1992.

DEC had been losing ground to newer 32-bit personal computers. Simplifying to three executives: one wanted to stay a systems company (“Boxes”), one wanted to build on customer relationships (“Solutions”), and one wanted to focus on semiconductors (“Chips”) ¹. The CEO, Ken Olsen, asked the group to reach consensus. It could not. A majority preferred Solutions to Boxes, a majority preferred Boxes to Chips, and a majority preferred Chips to Solutions, the cycle known as Condorcet's paradox ¹. The group settled on a statement: “DEC is committed to providing high-quality products and services and being a leader in data processing” ¹.

Rumelt calls that statement a political outcome, not a strategy ¹. Olsen was replaced in June 1992 by Robert Palmer, who chose Chips, five years too late in Rumelt's judgment; in 1998 Compaq acquired DEC ¹. The lesson for anyone running a strategy process: consensus among equally powerful executives with incompatible views can produce fluff by arithmetic. Someone has to own the choice.

Ask a divided room for consensus and you may get fluff, by arithmetic.

The second root is the template: fill in a vision, a mission, values and some goals labeled strategies ¹. A template has boxes for everything except the diagnosis, so a team can complete it without ever naming the obstacle. If your planning process hands out a template, add a first box, above everything else, labeled “What is the challenge, and why does it exist?”

The case for ambition, and the limits of the test

The strongest objection is that this checklist is a recipe for timid strategy. Stretch goals motivate people. Chad Logan was quoting Jack Welch: “We have found that by reaching for what appears to be the impossible, we often actually do the impossible” ¹. A strategist who insists on diagnosis first can sound like someone who would rather explain the problem than beat it.

Rumelt's answer is in the same passage. Logan's reading of Welch was selective; Welch also said, “If you don't have a competitive advantage, don't compete” ¹. A leader can ask for one last push, but the leader's job is also to create the conditions that make the push effective ¹. The checklist has no quarrel with ambition. It objects to ambition standing in for the explanation of how.

There is a second, subtler objection, and it is fair: the kernel can be gamed. A team that already knows what it wants to do can write a diagnosis shaped to justify it. Three guards help. Ask the diagnosis to explain the past, as Harvester's plan could not explain its margin. Ask what evidence would prove it wrong, and go and look. And have someone outside the sponsoring team draft a rival diagnosis. If two diagnoses fit the facts equally well, you have found the real strategic question, and that is worth more than either answer.

Finally, at portfolio scale the checklist applies differently. A corporate strategy and each business unit's strategy need their own kernel, because each faces a different obstacle. The corporate, business and functional strategy page covers what each level has to diagnose.

Monday morning

Print your company's strategy. Take one highlighter and mark only the sentences that explain why things are the way they are. Count them.

If the count is zero, you know what the next strategy conversation should be about, and it is not the action list. If the count is high but the sentences name ten problems, pick the one or two that, if solved, would make the others easier. Rumelt describes good strategy as focusing energy and resources on one or a very few pivotal objectives whose accomplishment leads to a cascade of favorable outcomes ¹.

The city with 178 action items did not need a 179th. It needed a sentence about what was wrong. A strategy starts there or it does not start, and the plan to create a plan was, in its way, an honest admission of that.

Questions

What is the kernel of good strategy?

Richard Rumelt's kernel has three parts: a diagnosis that explains the nature of the challenge and identifies the critical aspects of the situation, a guiding policy that sets an overall approach to overcoming the obstacles, and coherent actions that are coordinated with one another to carry out the policy. A document missing the diagnosis has no basis for choosing its policy or ranking its actions.

What are the four hallmarks of bad strategy?

In Rumelt's 2011 McKinsey Quarterly article they are failure to face the challenge, mistaking goals for strategy, bad strategic objectives and fluff. Bad objectives come in two forms: a long, unprioritized list of things to do, and blue-sky objectives that restate the desired outcome without any idea how to reach it. Fluff is buzzwords restating the obvious.

Is a stretch goal bad strategy?

Not by itself. A stretch goal becomes bad strategy when it stands in for the strategy, with no account of the obstacle or the approach that will overcome it. Rumelt's example is a CEO whose plan was 20% growth and 20% margin, with the will to win as the method. Keep the goal, and put a diagnosis and guiding policy beside it.

Why do organizations produce bad strategy?

Rumelt names two roots: the inability to choose, when powerful executives with incompatible views are asked for consensus and settle on a vague statement, and template-style planning, which supplies boxes for vision, mission, values and goals but none for the diagnosis. Both let a team finish a strategy document without ever naming the obstacle.

Where to go next

• What Strategy Is, and Five Things It Gets Mistaken For
• How Strategists Think: Hypothesis-Driven Problem Solving
• Strategy vs. Tactics: The One-Question Test
• What Is Strategy? A Complete Guide
• Strategic Plan
• Competitive Advantage