How to Score OKRs Honestly Without Grading People
How to score OKRs honestly: measure each key result against what happened, grade the goal and not the person, and treat an all-green quarter as a warning.
To score OKRs, compare each key result with what was actually achieved at the end of the quarter, without adjusting for how hard it was, how the quarter felt or who is in the room. Grade the goal, not the people, and keep scores out of pay and promotion.
The guide is clear that the long-term value of the method is a record of honest judgement. Scoring is the moment that record is made or spoiled, so it deserves more care than the ten minutes it usually gets.
How to score OKRs: the plain rule
The rule is simple to state and hard to follow. Take each key result and score it against what really happened. That is all.
Do not adjust for the difficulty of the quarter. Do not adjust for how the work felt, or for effort. Do not adjust because the owner is in the room, or because they worked hard, or because a good result would be awkward to announce. Each of those adjustments feels reasonable in isolation, and together they turn the score into a story about the team rather than a measurement of the goal.
A score is only worth recording if two readers would reach the same one from the same facts. That is why the starting value matters, since you cannot score a movement you never measured. See how to set a key result baseline for how to make that possible before the quarter starts.
Scores are instruments, not verdicts
The guide's phrase is that scores are diagnostic instruments. A thermometer does not blame anyone for a fever. The same attitude should apply to a result: it tells you something about the goal, the estimate and the route, and it should be read as information.
The moment scores become verdicts on people, they become fiction. People protect themselves. They set safe targets, they negotiate the arithmetic and they explain away the awkward numbers. Every benefit the method promised in its first part then evaporates, including the memory of judgement that depends on honest records. The wider case for that memory is in what are OKRs.
Reading a score by the kind of goal
A score only means something once you know what kind of goal it belonged to. The guide separates two kinds, and the reading is different.
| Kind of goal | What it means | How to read a low score | How to read a very high score |
|---|---|---|---|
| Committed | You expect to deliver in full, and less is a real problem | A genuine problem to understand | As expected |
| Aspirational | Set deliberately beyond comfortable reach | Frequently a good outcome and should be said so plainly | Suggests the aim was too low |
This is why labelling matters at the start. Without a label, nobody can tell whether a partial result is a success or a failure, and the argument about interpretation fills the gap. Agree what a good score looks like for each type before the quarter begins.
The all-green warning
A high score across everything feels like a triumph. The guide asks you to treat it as a prompt for a conversation about ambition instead of a celebration.
When every goal is comfortably on track every week and finishes strongly, something is wrong. There are three possible explanations, and the guide lists all of them.
- The goals were set inside what was going to happen anyway.
- People are managing the report rather than the work.
- Bad news is not travelling.
All three are serious, and all three look, from a dashboard, exactly like success. So the correct response to a uniformly green board is not satisfaction but an uncomfortable question. Ask which of the three applies, and look for evidence. The weekly signal that would have caught it is described in how to run a weekly OKR check in.
Keep scores away from pay
Everything above depends on one decision made before the quarter, and it is the most important decision in the process. Attainment must not feed reviews, bonuses or promotion cases.
The guide explains the reason in terms of incentives rather than character. When the score affects someone's pay, everyone sensibly sets targets they are sure of hitting, hides bad news until the last moment and argues about the scoring. That is not cynicism. It is the only reasonable response. And it damages both sides at once: the ambitious goal setting disappears, and the performance system is built on numbers that everyone has quietly tuned.
The guide adds a harder instruction. Say it out loud when you introduce the system, then hold the line at the first moment it is inconvenient, because that is when everyone finds out whether it was true. One instance of a missed goal quietly counting against someone will convert the whole company to conservative target setting within a single cycle, and you will not get that back.
A scoring session in order
Hold the session as a short, structured conversation, and keep the order so that the facts come before the feelings.
- Read the baseline and the target for each key result.
- Read what was actually achieved, from the agreed source.
- Record the score with no adjustment.
- Note the kind of goal, committed or aspirational, and read the score accordingly.
- Discuss low scores on aspirational goals as information, not failure.
- If everything is green, ask the ambition question before anything else.
The guide's end-of-quarter checklist covers the same ground: every key result scored against what actually happened, low scores on aspirational goals discussed as information, and an all-green result triggering a conversation about ambition.
Common ways scoring goes wrong
The method does not fail on arithmetic. It fails on behaviour around the arithmetic. These are the patterns to watch for on the day.
- Softening. A result that missed is described as close, or as good given the circumstances. The score is nudged up to match the description.
- Re-reading the target. The wording of the key result is quietly reinterpreted so the actual figure fits. A written baseline and target stop this.
- Scoring effort. The team worked very hard, so the number is lifted. Effort is real and deserves thanks, but it is not the thing being measured.
- Scoring the person. The conversation drifts from what happened to who is responsible. At that point the room has stopped looking at the goal.
Each of these has the same root. Somebody in the room fears what the score will be used for. If you hear them, the fix is not to scold the speaker but to check whether the separation from pay and reviews is really believed, and to restate it.
Scoring when the goal was declared dead
Sometimes a key result was abandoned mid-quarter, with a written reason. That is a success of the process, not a stain on the record. Do not quietly drop it from the list or score it as though it had been pursued to the end. Record that it was declared dead, when, and why.
That written reason is described in the guide as worth more in the retrospective than a completed goal would have been, because it shows what you learned and how early you were willing to say so.
What to do with the scores afterwards
Scoring is not the end of the cycle. It is the input to the retrospective, where you ask what you set out to change, whether it changed, where your estimates were wrong and in which direction. Without that step the scores are only a number on a page, and companies that skip it run the method for years without improving at setting goals.
The four questions and how to run the session are in how to run an OKR retrospective. If you find that scores keep coming back suspiciously high or oddly consistent, the patterns behind that are catalogued in OKR mistakes founders make.
Why generous scoring is the mistake that costs most
Scoring generously protects feelings for one afternoon. In exchange, it destroys the institutional memory that was the whole long-term benefit. The guide puts it plainly: honest scores are the asset, and good scores are not.
Consider what you lose. A year of inflated scores cannot tell you whether you overestimate, which bets worked or how long change takes in your business. It tells you only that everyone was kind. That is pleasant and useless, and a new joiner who reads the record learns nothing about how the company thinks.
Frequently asked questions
Should I adjust a score for a difficult quarter?
No. Score against what was actually achieved. Difficulty belongs in the retrospective discussion, where it can be learned from, and not in the number.
Is a low score a bad result?
Not on an aspirational goal. The guide says a low score there is frequently a good outcome and should be called that. On a committed goal, a low score is a real problem to understand.
Can OKR scores feed performance reviews?
The guide says no. Tying attainment to reviews, bonuses or promotion pushes people to sandbag targets and hide bad news, and it turns both the goals and the performance data into fiction.
What if every goal scores well?
Treat it as a warning and ask about ambition. The goals may have been set inside what would have happened anyway.
The full method, with the checklists, is in The OKR Playbook, a free 32-page playbook.