Measure What Matters: How Google, Bono, and the Gates Foundation Rock the World with OKRs

The summary

On a fall day in 1999, John Doerr drove to a two-story building off the 101 freeway with a slide deck. Two months earlier he had put $11.8 million into 12 percent of a company run by two Stanford dropouts, the biggest bet of his nineteen years as a venture capitalist. About thirty Googlers gathered around the ping-pong table that doubled as their boardroom. Sergey Brin’s verdict after ninety minutes: “Well, we need to have some organizing principle. We don’t have one, and this might as well be it.”

An objective is a direction, a key result is a milestone

The system fits in two lines. An objective is what you want to achieve; key results are how you’ll know you got there. Andy Grove’s own example, from a seminar Doerr sat in on as a 1975 intern: the objective is “dominate the mid-range microcomputer component business”; the quarter’s key result is “win ten new designs for the 8085.” One is a direction you can argue about for years. The other you settle by Friday. “Did I do that or did I not do it?” Grove said. “Yes? No? Simple. No judgments in it.”

Grove built on Peter Drucker’s management by objectives and repaired what had killed MBOs — goals planned centrally, refreshed once a year, hardened into context-free KPIs, and worst of all wired to bonuses. If a failed stretch costs you money, you stop stretching. His rules ran the other way:

• Three to five objectives per cycle, five or fewer key results each.
• Roughly half of all goals set from the bottom up.
• Anything can be revised or dropped mid-cycle when the ground shifts.
• Goals stay out of the bonus math — “a stopwatch in his own hand.”

Focus is mostly a list of things you refuse to do

Ask eleven thousand senior executives to name their company’s top priorities and a majority can’t; only half can name even one. So the first job is subtraction, and the danger is subtracting the wrong thing. Ford’s Pinto had two loud numbers, under 2,000 pounds and under $2,000, enforced by Lee Iacocca with an iron hand. When a crash test showed that a one-dollar piece of plastic stopped the gas tank puncturing, it was thrown out as extra cost and extra weight. Safety appeared nowhere in the car’s stated objectives. Hundreds died in rear-end collisions, and Ford recalled 1.5 million cars in 1978.

Grove’s fix is to pair key results, measuring effect and counter-effect: three new features paired with fewer than five bugs each, ten sales calls paired with two new orders. Quantity alone always finds a way to eat quality.

Operation Crush, and what a shared goal buys you

In November 1979, a district sales manager named Don Buckout fired off a desperate eight-page telex: Intel’s 8086 was losing to Motorola’s 68000. Within a week the executive staff had met; within two, a task force had a plan; before Christmas a nine-part program was approved. Jim Lally’s rallying cry needed no polish: “We have to kill Motorola, that’s the name of the game.”

Not one product was modified. What changed was the pitch — from ease of use to long-term systems and support, from selling programmers to selling CEOs — and the fact that each of the nine projects became a corporate key result, cascaded down to deadlines like “Deliver Rev 2.3 masks to fab on April 9.” A near-billion-dollar company turned on a dime. By 1986 the 8086 held 85 percent of the 16-bit market, and a bargain variant sat inside the first IBM PC. Intel graded that quarter’s corporate OKR at 0.625, and the board called it respectable, because they knew how aggressively the targets had been set.

Goals you are supposed to miss

Google sorts OKRs into two baskets. Committed goals get delivered in full. Aspirational goals are pitched so that 60 to 70 percent attainment counts as success. Gmail’s team debated offering 100 megabytes of storage and shipped a full gigabyte, up to five hundred times the competition. Sundar Pichai set Chrome’s 2008 target at 20 million weekly active users from a standing start and missed; in 2009 he aimed at 50 million and finished at 38. For 2010 he proposed 100 million, Larry Page pushed for more, and they settled on 111 million — reached that autumn after a passive alert to lapsed users moved the count from 87 to 107 million in weeks.

YouTube’s version was a billion hours of daily watch time, a 10x goal set in 2012 for the end of 2016. First Cristos Goodrow had to change the company metric from views to watch time, an argument that took six months and cost ad revenue up front. Then his team ground out roughly 150 improvements in 2016 alone, some worth 0.2 percent each, and hit the number ahead of schedule.

Scores can’t hold a conversation

The second half of the book is about what OKRs don’t do. Annual reviews eat 7.5 hours of manager time per direct report, and only 6 percent of HR leaders think they’re worth it. Adobe was burning 80,000 manager hours a year on a process that spiked voluntary attrition every February. Donna Morris scrapped it for “Check-in”: quarterly goals, feedback at least every six weeks, compensation handled separately. Attrition fell. Doerr calls this half CFRs — conversation, feedback, recognition — and the separation from pay is the load-bearing part. Google keeps OKR scores to a third or less of a performance rating and wipes the raw numbers after each cycle. Tie goals to bonuses and people quietly set goals they can hit.

The bottom line

Pick a handful of goals, attach numbers, let everyone see them, check in weekly, and aim high enough that you expect to fall short of some — that’s the whole method, and Intel and Chrome show it holding up under real pressure. Know what else you’re buying, though: half the book is Doerr’s portfolio companies taking turns at the microphone, and in most of those chapters OKRs accompanied the success rather than caused it. Read it if you lead a team and your quarterly plan is really a wish list, and take Bono’s warning seriously — the framework can make you too organized. “If everything’s at green, you failed.”