OKRs: goals that grade themselves
Grove's original Intel system and what Doerr carried to Google — why 0.7 is a success, why OKRs must never touch compensation, and the committed-versus-aspirational split most teams miss.
The theory in one paragraph
An OKR pairs an Objective — qualitative, directional, memorable — with three to five Key Results: the numbers that prove you arrived. Andy Grove's insight at Intel was to separate two questions managers habitually blur: 'where do I want to go?' and 'how will I know I'm getting there?' Answered separately, they let knowledge workers manage themselves against measurable outcomes instead of supervised task lists. John Doerr, who learned the system as a young Intel engineer, presented it to a 40-person Google in 1999; Google has run on it every quarter since. The mechanics are almost trivially simple — which is precisely why the discipline around them (honest scoring, short cadence, no link to pay) is where the entire framework actually lives.
How it works
The mechanics — as Andy Grove defined them, not the folklore version.
Objectives are direction; key results are evidence
The objective is qualitative on purpose — 'make onboarding feel effortless' — because its job is to be remembered and to motivate. The key results carry all the rigor: 'activation rate from 20% to 35%', 'median time-to-first-value under 10 minutes'. Grove's test still applies: at quarter's end, a stranger should be able to score each key result 0 to 1.0 with no debate. If scoring requires a discussion, you wrote a task or a hope, not a key result.
Quarterly cadence, brutally honest grading
OKRs are set at quarter start and scored at quarter end on a 0–1.0 scale — then mostly thrown away and rewritten, which is what makes them cheap to be ambitious with. The scores exist to calibrate, not to punish: a team averaging 1.0 isn't excellent, it's sandbagging, and a team averaging 0.2 set fantasies. The system self-corrects only if the grading stays honest, which is why what the scores are used for matters more than the scores.
Committed versus aspirational — Google's load-bearing split
Committed OKRs are promises — ship the feature, hit the SLA — where 1.0 is expected and a miss triggers a real postmortem. Aspirational OKRs are moonshots where ~0.7 is success and 1.0 means you aimed too low. Collapsing the two is the most common OKR failure: treat moonshots as commitments and you burn the team; treat commitments as stretch goals and nothing ships on time.
The moment OKRs touch pay, they die
Both Grove and Doerr are explicit on this and almost every adopter ignores it: OKR scores must not feed bonuses or performance reviews. The system's value is that it makes ambition legible — people write down goals they might miss. Attach money to the score and every rational employee sandbags, 0.7-as-success inverts into 0.7-as-pay-cut, and your goal system quietly becomes a negotiation system. Doerr's line is the whole argument: OKRs are a compass, not a dashboard for compensation.
The person behind it
Andy Grove & John Doerr
Intel's legendary CEO · the venture capitalist who carried his system to Google
Grove built OKRs at Intel in the 1970s as 'iMBO' — his repair of Peter Drucker's Management by Objectives, adding a faster cadence, self-set measurable results, and a firewall between goals and performance reviews. Doerr absorbed the system in Grove's Intel classes, and as a Kleiner Perkins investor pitched it to Larry Page and Sergey Brin in 1999, when Google was 40 people in a converted house. Measure What Matters, two decades later, is Doerr's account of why it stuck — and his warning about every way companies deploy it wrong.
Lineage — Grove's repair of Peter Drucker's Management by Objectives — shorter cadence, self-set measurable results, decoupled from pay
How to apply it this week
Each step maps to a field in the OKRs tool — finishing the read means finishing the work.
Write one to three objectives — qualitative, time-bound, worth wanting
Each objective is a direction the quarter should bend toward, phrased so the team can recite it. If it contains a number, it's a key result that snuck upstairs. If nobody would be excited to hit it, it's a chore, not an objective.
OKRs · objectivesAttach three to five key results per objective — outcomes, not tasks
'Increase activation from 20% to 35%' is a key result; 'redesign onboarding' is a task that might cause it. The from–to structure matters: it forces you to know the baseline, which is where most first-quarter OKRs quietly fall apart.
OKRs · key resultsLabel every OKR committed or aspirational
Decide out loud which promises are promises. Committed means 1.0 or a postmortem; aspirational means 0.7 is a win. Leaving the label off lets everyone assume the interpretation that flatters them — which defeats the point of writing goals down.
Score at quarter end, in the open, and calibrate
Grade each key result 0–1.0 from the data, discuss the surprises, then reset. Straight 1.0s mean next quarter's targets move up; a 0.3 gets a why, not a blame. The scoring meeting is the system — skip it twice and you're back to a wishlist in a spreadsheet.
OKRs · scoringKeep the firewall: scores never feed compensation
Say it in writing when you introduce OKRs, and honor it forever. Performance reviews can discuss what someone attempted and learned; the moment the numeric score sets pay, ambition exits the goal-setting process and never comes back.
OKRs · quarterly reset
Build it, don't just read it
The steps above are the OKRs tool's structure. Open it and work through them with your own startup — it's free, no card required.
Free tool · free account · no card required
See it in the wild
Teardowns from our benchmarks library where this framework is doing real work.
Frequently asked questions
What does OKR stand for and how does it work?
OKR stands for Objectives and Key Results. An objective is a qualitative, inspiring goal ('own the developer onboarding experience'); each objective gets three to five key results — measurable outcomes with numbers and deadlines that prove the objective was reached. Teams set them quarterly, score each key result from 0 to 1.0 at quarter's end, and rewrite them fresh — the short cadence is what keeps them honest.
Why is 0.7 considered a good OKR score?
Because aspirational OKRs are deliberately set beyond what the team knows how to achieve — Google's convention is that landing around 70% of an ambitious target is success, while consistently scoring 1.0 means the targets were sandbagged. The 0.7 norm only applies to aspirational OKRs, though: committed OKRs are promises where anything under 1.0 triggers a postmortem.
Should OKRs be tied to compensation or performance reviews?
No — this is the rule both Grove and Doerr state most emphatically and companies break most often. If OKR scores determine bonuses, employees rationally set targets they're certain to hit, ambition drains out of the system, and the 0.7-is-success convention becomes impossible. Keep goal-setting and pay conversations structurally separate; reviews can discuss what someone took on and learned, never the raw score.
What is the difference between committed and aspirational OKRs?
Committed OKRs are obligations — launches, SLAs, revenue floors — where the expected score is 1.0 and a miss requires explanation and a plan. Aspirational OKRs (Google calls them moonshots) describe where you'd love to be, where ~0.7 is a strong result and 1.0 suggests you under-aimed. Labeling each OKR explicitly is essential: unlabeled goals get treated as commitments by managers and as stretch goals by teams, simultaneously.
What is the difference between OKRs and KPIs?
KPIs are ongoing health metrics — uptime, CAC, churn — that you monitor continuously and never 'finish'. OKRs are time-boxed change goals: where you're deliberately pushing this quarter and how you'll know you succeeded. They cooperate rather than compete — a deteriorating KPI often becomes next quarter's OKR, and a good key result frequently targets moving a KPI from one level to another.
Related frameworks
North Star Metric
Sean Ellis · 2017
The Lean Startup
Eric Ries · 2011
Business Model Canvas
Alexander Osterwalder · 2005
Sources
Independent educational summary written by StartupKit from public sources. OKRs is the work of Andy Grove & John Doerr; this page is not affiliated with or endorsed by the author.