In 1992 Digital Equipment Corporation was losing ground. It had helped lead the minicomputer revolution, and for several years newer 32-bit personal computers had been eating into its business. Richard Rumelt, a strategy professor at UCLA, sat in while its senior executives argued about what to do.
Rumelt later simplified the room to three people. “Alec” wanted DEC to keep building integrated systems: Boxes. “Beverly” wanted to build on its customer relationships and sell Solutions. “Craig” wanted to put the company’s resources into semiconductors: Chips. Each new path would remake the company, and in Rumelt’s view nobody could do both at once.
Ken Olsen, the founder and chief executive, asked the group to reach a consensus. A majority preferred Solutions to Boxes. A majority preferred Boxes to Chips. And a majority preferred Chips to Solutions. Whichever plan they picked, a majority preferred something else.
So they settled on a sentence everyone could accept: “DEC is committed to providing high-quality products and services and being a leader in data processing.”
It asked none of them to give anything up.
A sentence with nothing in it
Rumelt called it “a political outcome,” not a strategy. In a 2011 article for McKinsey Quarterly he gave the pattern a name: bad strategy. A goal, a slogan or a wish list can pass for a strategy. None of them chooses anything.
He boiled its hallmarks down to four. The first is fluff, buzzwords that dress up the obvious. He quotes a retail bank’s internal memo: “Our fundamental strategy is one of customer-centric intermediation.”“Remove the fluff,” he wrote, “and you learn that the bank’s fundamental strategy is being a bank.”
The second is failing to face the challenge. In 1979 International Harvester’s planners produced a thick corporate strategic plan built on more market share and lower costs in every business. By Rumelt’s account it never mentioned the company’s grossly inefficient plants, or that it had the worst labor relations in American industry. After a disastrous six-month strike, the company began to collapse.
The third is mistaking goals for strategy. Bad strategy, he writes, is like a quarterback whose only advice to his teammates is “let’s win.” The fourth is bad strategic objectives: a long list of things to do, or targets as hard to reach as the problem they were meant to solve.
“Arrive safely and on time” is a wish. A route names the bridge that’s out and which way you’ll go around it.
Why the room couldn’t choose
DEC’s executives weren’t short of ideas. They had three, and that was the trouble. “Strategy involves focus and, therefore, choice,” Rumelt wrote. “And choice means setting aside some goals in favor of others.” Asking for consensus made setting anything aside impossible, because for each plan a majority preferred another one.
His second root cause is the planning template: a vision, a mission, some values, then a list of aspirations labeled strategies. Each box is easy to fill and none of them asks what is in the way.
Rumelt isn’t against goals or visions, which he calls “an important part of human life,” but no substitute for “the hard work of strategy.” Nor does choosing make a choice right. Pick the wrong obstacle and you can focus hard on the wrong thing. Rumelt first used the term at a 2007 Washington seminar on national-security strategy, and the DEC meeting, the bank memo and the Harvester plan all come to us through his accounts.
The roadmap with room for everyone
Imagine you run a twelve-person product team. One lead wants the quarter spent on crashes, which come up again and again in cancellation tickets. Another wants the reporting feature a large prospect has asked for. A third wants to rebuild sign-up.
The draft on your desk says: “Deliver a reliable, delightful experience for every customer.” All three leads approve it, because all three plans fit inside it. Three months later you have three projects, each half done.
The other draft takes a side. Cancellations are the problem, and crashes are the main reason customers give, so this quarter is crashes. Reporting waits, and someone tells the prospect. Sign-up waits too.
The prospect may walk. That cost was there in the first draft as well. The first draft just didn’t say so.
Asking a model to merge the memos
Now imagine you paste the three leads’ memos into a chat model and ask for “one strategy that reflects the whole team.” That’s Olsen’s request. One way to satisfy it is to do what DEC’s executives did and write a line broad enough for all three plans to fit inside. The model isn’t defending any team’s turf. The blur would come from what you asked for.
Ask for a choice instead. Have it set out the three plans as options that exclude each other, and for each one name the problem it assumes is the main one, the evidence in the memos for that, and what the team would stop doing. Then you choose.
Two checks are easy to run. Swap your company’s name for a competitor’s and mark every sentence that is still true; those are the first suspects for fluff. Then run the request a few more times with the memos in a different order. If the recommended option keeps changing, take that as a sign to check each option’s reasons against your own tickets and numbers, not as an answer.
Five years
In July 1992 Olsen announced he would retire, and DEC named Robert Palmer, a vice president of manufacturing, to succeed him. By Rumelt’s account Palmer, who had headed semiconductor engineering, made it clear the strategy would be Chips. “One point of view had finally won,” Rumelt wrote. “But by then it was five years too late.”
In June 1998, Compaq completed its acquisition of Digital Equipment.