In one sentence: A venture capitalist explains the goal-setting system he learned from Andy Grove at Intel and taught to Google in 1999, in which a qualitative objective is paired with three to five measurable key results, scored in public every quarter, and surrounded by continuous conversation rather than an annual review.
At a Glance
Author: John Doerr
First published: 2018 (Portfolio)
Category: Business / Management
Length: about 320 pages, about 72,000 words (Portfolio hardcover)
ISBN-13: 978-0-525-53622-2 (Portfolio hardcover)
Summary reading time: about 12 minutes
Book reading time: about 5 hours
Notable adaptations: no film or series, though the book turned OKRs into a standard corporate practice and spawned an industry of software and consultancies, and Doerr followed it with Speed and Scale in 2021
Content note: roughly half of this book is written by other people. Doerr’s chapters alternate with first-person case studies by the founders and executives whose companies are praised, among them Bill Gates, Bono, Sundar Pichai and Susan Wojcicki, so those passages are testimony, not analysis. Doerr is a venture capitalist whose firm invested in several of the companies featured, and he put 11.8 million dollars into Google before teaching it the method. Every figure predates 2018, and the Resources templates survive as text rather than usable forms.
Doerr joined Intel as an intern in 1975 and spent five years under Andy Grove, who had begun running the company on objectives and key results, or OKRs, in 1971. The lineage runs back to Peter Drucker’s management by objectives in 1954, which Grove made quarterly, public, partly bottom-up and mostly separate from pay. Doerr took it to Kleiner Perkins in 1980 and, in the fall of 1999, to a company of about thirty-five people called Google. The book repeats that arrangement: Doerr introduces, a guest writes the case, Doerr draws the lesson.
Read it if you are introducing a goal system and want the founding text and its vocabulary. Skip it if you want evidence that the method works, or a candid account of its costs.
The Big Idea
The framework fits in a paragraph, which is both its strength and the reason a summary can stand in for the book. An objective is what you want to achieve, qualitative and inspirational. Key results are how you know you got there: specific, time-bound, aggressive yet realistic, and above all measurable. Three to five per objective is the guidance, and Doerr’s test is strict. If you complete every key result and still have not achieved the objective, you did not write an objective. A formula he credits to Intel’s Bill Davidow joins the halves: we will achieve a certain objective as measured by the following key results.
Two design choices do most of the work, and both are the parts organizations quietly drop. The first is publication: goals at Intel and Google are visible company-wide, which makes alignment visible rather than asserted. The second is scoring, which runs by different rules for the two classes of goal. A committed OKR is a promise expected to come in at 1.0. An aspirational OKR is set beyond reach, with an average near 0.7 counting as success, which is why Doerr can say Google fails at roughly 40 percent of its aspirational goals by design.
Part Two argues that measurement alone is half a system. Scores say where a team stands and nothing about why, so Doerr pairs OKRs with what he calls CFRs: conversations, feedback and recognition. From that comes his second and more contentious prescription, that the annual review be abolished and goals divorced from pay, since goals used to set pay invite sandbagging.
Key Ideas
1. Write the objective first, then the numbers that would prove it
Here is the method, with an example. State the objective as an outcome rather than the work you plan, in language a new hire would find motivating: make new users successful in their first week. Then ask what would have to be measurably true if that had happened, and write three key results, each with a baseline, a target and a date. Raise seven-day activation from 34 percent to 50 percent by March 31. Cut median time to a user’s first completed project from 48 hours to 6 hours. Hold support tickets at or below 12 per 100 new accounts. The third protects the first two, since Doerr advises pairing a quantity measure with a quality measure. Read it back as one sentence using the formula: if all three numbers could land while the objective stayed unmet, the key results are wrong. Then grade each key result from 0 to 1.0 against its target, average them for the objective, and read the total on the three-band scale in Google’s playbook: 0.7 to 1.0 green, 0.4 to 0.6 yellow, 0.0 to 0.3 red. Doerr grades Intel’s 1980 corporate OKR at 0.6, 1.0, 0 and 0.9, an average of 0.625. Add a few sentences of self-assessment, since the score starts the conversation rather than ending it.
2. Focus means choosing a few things and saying so out loud
The first superpower is focus and commit to priorities, and the hygiene rules are blunt: three to five objectives a cycle, five or fewer key results each, nothing merely business as usual. The constraint is the point: fifteen priorities tell a team nothing about what to drop when the quarter gets tight. Leaders must model the choice, since an objective the executive team visibly ignores teaches everyone that the system is theater. Remind’s cofounder admits the commonest failure in his own chapter, setting seven or eight objectives for a team with capacity for two.
3. Alignment works sideways and upward, not only down
Superpower two is align and connect, and the argument against pure cascading is the better half of it. Goals pushed down a hierarchy lose time at every level, link only vertically, and leave the people closest to the work nothing to offer but compliance. Doerr’s prescription is that about half of all goals start from the bottom up, with teams writing their own objectives and choosing which company objective theirs supports. Transparency does the connecting, letting two teams find a dependency without a meeting.
4. Tracking is the dull habit that makes the rest work
Tracking for accountability, the third superpower, governs the weeks between setting goals and scoring them, which is the stage most teams skip. It means a regular check allowing four answers for a drifting goal: continue it, update the target or timing, start something new, or stop it. Stopping is hardest, and it is where the line about goals serving purpose earns its place: a goal overtaken by events is information, not a verdict.
5. Committed goals and stretch goals are different animals
The fourth superpower is stretch for amazing, and its central distinction is the book’s most useful contribution. Committed goals are forecasts the organization will resource and deliver in full. Aspirational goals are set at something like ten times current performance and expected to land around 60 to 70 percent. Mixing them destroys both, since a stretch target scored as a promise teaches people to stop stretching. The showcase is YouTube, in a chapter by Susan Wojcicki and Cristos Goodrow, the executives responsible. By their account a target of one billion hours of daily watch time was set in late 2012, when the figure stood at roughly a tenth of that, and reached in the fall of 2016. Their account carries the honest details too, including the roughly 150 improvements of about 0.2 percent each in the final year, and watch time’s later replacement, because a metric has a shelf life.
6. Conversations, feedback and recognition replace the review
The case against what CFRs replace is the stronger half of Part Two. Doerr cites Deloitte’s finding that annual reviews consume about 7.5 manager hours per direct report, and that 12 percent of human resources leaders consider them highly effective while 6 percent think them worth the time. In their place he wants frequent one-on-ones built on five standing questions about work, goals, obstacles, support and growth, plus peer feedback and visible recognition. The pay question draws the most hedging: he separates the compensation conversation from the development one, then concedes that goals and rewards should never be completely severed.
7. Culture is the medium, not the output
The closing argument is that OKRs cannot work in a vacuum. He assembles the chapter from other people’s research, mapping Project Aristotle and the Progress Principle onto his own two instruments. The company that proves the point is Lumeris, whose chief human resources officer writes that goal setting failed for three cycles as window dressing, and worked only once autocratic leaders had been moved out.
Context and Analysis
What holds up is the framework. Separating a qualitative objective from the measurable results that would prove it is a real discipline, and it stops teams reporting activity as progress. The system is cheap, legible and portable, which is most of why it spread, and the committed-versus-aspirational split is a genuine contribution. Doerr’s own caveat is more honest than much of what follows it: the method is no silver bullet and cannot substitute for judgment, leadership or workplace culture.
The evidence, though, is promotional by construction. Every company shown succeeded, the cases are written by the executives responsible, and Doerr’s firm had money in several of them, so no company appears that ran the method faithfully and failed, which is the comparison the argument needs. Causation is asserted: the claim that Operation Crush would not have happened without OKRs is indistinguishable from the effect of funding, talent or a competitor’s misstep. Outcomes arrive in the wrong currency too, as acquisitions and user counts rather than goal-attainment rates. The definitions have a circular edge, since a goal that misbehaves is declared not to be a real OKR, and 0.7 counts as success for one class of goal while anything under 1.0 is failure for another. Weakest is the half of the prescription that rests on assertion. The CFR chapter’s figures describe the failings of annual reviews rather than the benefits of its replacement, and Adobe’s headline result, that voluntary attrition dropped sharply, arrives by Adobe’s own account with no figure, baseline or control. Decoupling from pay is hedged to near-meaninglessness, since goal scores remain a third or less of a Google rating and the playbook recommends reassigning headcount away from sandbaggers. The culture chapter relabels other people’s research, its striking figures are self-reported, and Lumeris shows culture had to precede the method. The Goals Gone Wild research on how stretch targets narrow focus and encourage unethical behavior gets a rebuttal rather than an engagement.
Two books make useful company. Trillion Dollar Coach is about Bill Campbell, who appears throughout this one and to whom Doerr dedicates it. It is the better account of the coaching relationship CFRs try to systematize, because Campbell’s method was personal and unscalable, precisely what a framework cannot manufacture. Simon Sinek’s The Infinite Game supplies the sharpest objection, that fixed targets distort behavior in games with no finish line, which anyone adopting a quarterly scoring system should hear.
What It Teaches
Six things a reader can act on this week.
1. Write the objective as an outcome in plain language, then ask what would have to be measurably true if it had been achieved. 2. Give every key result a baseline, a target and a date, and keep the count to three to five per objective. 3. Pair a volume measure with a quality measure, so hitting the number cannot damage what it stands for. 4. Decide before the quarter starts whether each goal is committed or aspirational, and score the first against 1.0 and the second against roughly 0.7. 5. Publish the goal set where the team can read it, and grade each key result from 0 to 1.0 at the cycle’s end. 6. Check in mid-cycle and allow four answers for a drifting goal, continue, update, start and stop.
Memorable Lines
“Ideas are easy. Execution is everything.” (Doerr)
“Objectives are the stuff of inspiration and far horizons. Key results are more earthbound and metric-driven.” (Doerr)
“Innovation tends to dwell less at the center of an organization than at its edges.” (Doerr)
“Our goals are servants to our purpose, not the other way around.” (Doerr)
“Annual performance reviews are costly, exhausting, and mostly futile.” (Doerr)
“Peak performance is the product of collaboration and accountability.” (Doerr)
Should You Read the Full Book?
Verdict: Summary is enough
The method compresses to a page or two, and you have just read it. What fills the 320 pages is testimony, thirteen case studies by people with a stake in how their companies are described, assembled by an investor in several of them. If you do open it, read Doerr’s four superpower chapters for the hygiene rules and failure modes, and the YouTube chapter for the book’s one fully worked stretch goal. Treat the rest as a case for the method rather than a test of it. The Measure What Matters book page has the full details and where to get a copy.