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

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

Author: John Doerr • Category: Business & Leadership

Doerr's argument is narrow: organizations rarely fail for lack of ideas, they fail because nothing turns an idea into a commitment someone can check. Objectives and Key Results is that translation layer. Drucker's Management by Objectives became Andy Grove's practice at Intel, and reached Google in 1999, when Doerr taught the template to Larry Page and Sergey Brin. The Gates Foundation's polio work and Bono's ONE campaign later ran on the same framework.

5 Pillars of Success

1. Separate intent from evidence The system fits in one sentence: "I will [Objective] as measured by [Key Results]." The Objective is qualitative — significant, concrete, action-oriented, inspirational. Key Results are quantitative — specific, time-bound, aggressive but achievable, and verifiable by someone other than the author. If a Key Result can be satisfied by an opinion, it is not one. Grove's two questions are the engine: What do you want to accomplish? How will you know you are successful?

2. Focus: three to five, and name what you are not doing Commitment comes through limitation. Cap Objectives at three to five per cycle at every level, and Key Results at three to five per Objective; everything off the list is explicitly not being done. OKRs describe outcomes, not activity: "ship the feature" is a task, while "the feature is used by a defined share of its target users" is a result. Intel's Operation Crush in 1980 is the extreme case. Facing Motorola in the 16-bit microprocessor market, Grove pointed the whole company at one objective — win the design-win battle — with a hard count of customer wins due by a fixed deadline. Everything that did not serve it was deprioritized.

3. Align by network, not by cascade Cascading — each layer subdividing the one above — is brittle and slow: every goal waits on a chain of approvals. Doerr's alternative is a network: publish every OKR and let teams link theirs to the company Objectives they serve, so dependencies surface instead of being managed by rank. Set a few top-level Objectives, then require teams to draft roughly half their own goals from the bottom up. Intuit ran exactly this mix company-wide, pairing top-down company OKRs with bottom-up team OKRs. At Google every goal is readable by every employee, including the chief executive's.

4. Track, grade, and keep pay out of it Weekly or biweekly check-ins review progress against the numbers; the cycle ends with a public 0.0–1.0 grade assigned by the team that wrote the OKRs. Label each OKR committed or aspirational as you write it: committed means 1.0 is expected and 0.8 is a problem worth explaining, while aspirational means an average near 0.7 is success. Never grade the two with one rule, and never tie OKRs to bonuses. The reason is behavioral — once the grade sets pay, people write safe goals, negotiate targets down, and manage the metric instead of the work. The companion shift is CFR — Conversations, Feedback, Recognition — replacing the annual review; Adobe swapped its review for frequent check-ins, and Accenture dropped annual ratings.

5. Stretch for amazing An aspirational goal does more than motivate: if it cannot be reached by working harder inside the current system, it exposes which part of the system is wrong. YouTube set one objective — one billion hours of watch time per day — and reached it in 2016; chasing watch time rather than clicks forced the recommendation system to optimize for whether viewers stayed and were satisfied, changing what the product, and eventually the company, optimized for.

Key Takeaway > "Ideas are easy. Execution is everything."

Practical Task 1. Write one Objective for the next quarter in a single sentence. If it could describe any quarter at any company, it is too vague to direct anything. 2. Apply Grove's second question three times and keep only answers a stranger could verify. "Improve onboarding" fails; "a new user completes their first summary within ten minutes of install" passes. 3. Cut to three Objectives and write down what you are not doing. If nothing was cut, the list describes the status quo, not priorities. 4. Give each Objective three to five Key Results and mark each committed or aspirational before work starts — the label decides how the final grade is read. 5. Publish every OKR where the team can see it, including leadership's, and link each team goal to a company Objective. 6. Book a weekly 30-minute check-in against the numbers and name the one person who runs it. This is the step most failed implementations skip. 7. Grade at cycle's end, 0.0 to 1.0, self-assessed and explained, with the grades kept out of pay and promotion decisions. 8. Pair every quantity metric with a quality counter-metric, so the target cannot be won by degrading the product.