OKR grading is scoring each key result from 0.0 to 1.0 at the end of a cycle: 0 means no progress, 1.0 means fully delivered, and around 0.7 is the target for a healthy stretch. An objective’s score is the average of its key results. The point isn’t the number, though: it’s the honest conversation the number forces.
Scoring is where OKRs either become a learning habit or quietly turn into a report card nobody trusts. This guide covers the 0.0–1.0 scale, how to score a key result and roll up an objective, the confidence check you run mid-cycle, and how to run the grading session itself. New to the framework? Start with the OKR guide or learn to write them in how to write OKRs →
The 0.0–1.0 scale
OKRs are scored on a scale from 0.0 to 1.0: the share of the target you reached. It helps to think in three bands rather than chasing a precise decimal.
Missed
Little real progress. Not a punishment, a prompt to ask what got in the way, and whether it was even the right goal.
Partial
You moved the needle but fell short. Common and healthy: most honest stretch goals land here some of the time.
Delivered
You hit, or nearly hit, an ambitious target. Around 0.7 is the sweet spot; a steady 1.0 usually means the goal was too safe.
Why 0.7 is the target, not 1.0
On a true OKR, a score around 0.7 is success, not failure, and that is the part which trips people up. OKRs are meant to be ambitious (a genuine stretch), so if you set them right, you shouldn’t fully hit every one. A team that scores 1.0 across the board almost certainly aimed too low.
If your scores are perfect every cycle, the problem usually isn’t that you’re a superteam: it’s that the targets were safe. Aim higher until landing at ~0.7 feels earned. And keep scores away from bonuses and ratings: the moment a number affects pay, people set goals they know they’ll hit.
This topic is lesson 15 of 17 in From Zero to Goals: Comparing cycles: trend beats a point. Read two quarters side by side and judge the trajectory, not the single score. You make the calls inside one realistic company, run the exercise on your own goals, and the course ends with a 15 question exam (12 to pass) and a certificate worth putting on your profile.
How to score a key result
A well-written key result scores itself, because it has a baseline and a target. The formula is just the fraction of the distance you covered:
Key-result score = (result − start) ÷ (target − start), capped at 1.0. So a key result to grow trial signups from 200 to 400 that lands at 350 scores (350 − 200) ÷ (400 − 200) = 0.75.
This is exactly why key results have to be measurable outcomes with a “from X to Y” shape. See how to write OKRs. A key result written as a task (“launch the redesign”) can’t be scored honestly: it’s either done or not, and everyone argues the number.
Worked examples: five key results, scored
The formula is one line, but four of these five cases trip teams up the first time they close a cycle. Work through them once and the rest of the scoring system stops being a debate.
0.75: grew as planned
Trial signups from 200 to 400, landed at 350. (350 − 200) ÷ (400 − 200) = 150 ÷ 200 = 0.75.
0.67: reduced churn
Monthly churn from 8% to 2%, landed at 4%. Distance is still start minus result over start minus target: (8 − 4) ÷ (8 − 2) = 4 ÷ 6 = 0.67. Never divide the two numbers: 8 ÷ 4 would read as 200%.
1.0: capped, and worth a question
Revenue from €40k to €60k, landed at €70k. (70 − 40) ÷ (60 − 40) = 1.5, capped at 1.0. Two 1.5s in a row means the targets, not the team, need adjusting.
0.0: the floor
Response time from 40 hours to 20, ended at 45. A negative fraction floors at 0.0. Scores never go below zero, so a bad quarter reads as 0.0, not −0.25.
0 or 1, and use it sparingly
“Ship the new billing page” has no scale: it lands or it doesn’t. Score it 1.0 or 0.0 and accept that it tells you nothing about how close you came. If most of your key results look like this, they’re tasks, not key results.
Always measure from the real baseline
If the metric was already at 250 when the cycle opened, the start is 250, not the 200 from last quarter’s plan. Scoring against a stale baseline is the most common way teams accidentally inflate a result.
How to score an objective
An objective doesn’t get its own separate grade: its score is the average of its key results. Three key results scoring 0.8, 0.6, and 0.4 give the objective a 0.6. Treat that as a health read of the objective overall (“mostly there,” “half there”), not a precise measurement. If one key result dominates the objective, that’s usually a sign the others weren’t pulling their weight and the OKR needed rebalancing. Numbers you watch for ongoing health rather than score at the close belong on a dashboard instead: that line is drawn in OKRs vs KPIs →
The rollup trap: unscored key results. Averaging only the key results someone got round to scoring quietly inflates the objective. An objective with three key results where one is scored 1.0 and two were never touched is not a 1.0: it’s a 0.33 if you count the untouched ones as zero, and an unknown if you’re honest. Decide the rule before the session and apply it every time:
Objective score = sum of all key-result scores ÷ the number of key results the objective has, not the number that got scored. An unscored key result counts as 0.0, because nobody can show it moved. If that feels harsh, the fix is to score it, not to leave it out of the divisor.
Both rules on this page (the distance formula and this divisor) are wired into our OKR score calculator, if you would rather type your numbers in than do the arithmetic by hand.
Confidence during the cycle vs. score at the close
Scoring isn’t only an end-of-cycle event. The best teams track a confidence read every week, so a key result heading for a 0.2 shows up in week 4 (while you can still do something about it), not on grading day.
During the cycle: confidence
At each check-in, the owner rates how likely the key result is to land: a quick 1-to-10, or on-track / at-risk / off-track. It’s a forecast, and a falling number is your early-warning system.
At the close: score
At the end you grade what actually happened, 0.0 to 1.0. It’s a result, not a prediction. Confidence tells you where to step in mid-cycle; the score tells you what to learn once the cycle’s done.
The OKR scorecard: one page you can actually read
An OKR scorecard is simply the whole cycle laid out on one page: every objective, its key results, the baseline and target, where the number landed, and the resulting score. It isn’t a separate methodology or a piece of software; it’s the artefact you bring to the grading session so nobody is scrolling through a document looking for last quarter’s numbers.
Six columns are enough, and a seventh is usually clutter:
Objective
The qualitative direction, repeated once per group of key results so the page reads top-down.
Key result
One line each, written as “from X to Y”: if it doesn’t fit that shape, it can’t be scored cleanly.
Start → target
Both ends, recorded when the cycle opened. Filling this in at the close is how baselines get rewritten.
Result
Where the number actually landed, with the date you read it. Two people reading a dashboard a week apart get different numbers.
Score
0.0–1.0 from the formula above. Unscored counts as 0.0 in the objective rollup.
What we learned
One sentence. The column everyone skips, and the only one still useful three months later.
Keep last cycle’s scorecard next to this one when you plan. A scoring system earns its keep across cycles, not within one: the pattern of a team landing 0.9 every quarter, or 0.3 every quarter, says far more than any single number on the page.
How to run an OKR grading session
At the close of the cycle, grade the OKRs together in one short session: it’s the natural end of your quarterly rhythm. Five moves keep it honest and useful, and for the meeting craft around them (agenda, who attends, keeping it blameless), see how to run a goal review meeting →
Pull the numbers before you meet
Gather the actual result for every key result ahead of the session, so you’re grading from data, not memory. Ten minutes of prep saves an hour of arguing.
Score each key result 0.0–1.0
Compare where you landed to the target: (result − start) ÷ (target − start), capped at 1.0. Measurable key results make this arithmetic; fuzzy ones make it a debate.
Roll up each objective
Average its key-result scores. Read the number as a health signal (“mostly there,” “half there”), not a precise grade.
Talk about the why, not just the number
For each objective, ask what worked, what blocked you, and whether the target was right. The score opens the conversation; it isn’t the conversation.
Write down what you’ll carry forward
Capture what to keep, drop, or reshape next cycle. Grading only pays off if the next quarter starts smarter than this one did.
Keep scoring honest
Two habits quietly wreck OKR scores. The first is sandbagging: setting targets you know you’ll crush so every score reads 1.0; the fix is to aim higher until ~0.7 feels earned. The second is tying scores to pay or ratings, which trains people to lowball; keep grading a learning tool, separate from performance reviews. For the full field guide across writing, running, and scoring, see common OKR mistakes →
In OKR mode, Celorly scores each key result from 0.0 to 1.0 at the close and rolls them into the objective automatically: grading is a few clicks, not a spreadsheet. Through the cycle, the Monitor screen surfaces the key results going quiet so you can act early, and scores stay where they belong: a learning tool, not a performance rating. See what Celorly does →
OKR grading and scoring: quick answers
How do you score an OKR?+
Score each key result from 0.0 to 1.0 by comparing where you landed against the target: (result − start) ÷ (target − start), capped at 1.0. The objective’s score is the average of its key results. Aim for around 0.7 on a genuine stretch: a consistent 1.0 usually means you set the bar too low.
What OKR score should you aim for?+
For a real stretch goal, about 0.7: it means strong progress on something ambitious. Scores near 1.0 every cycle suggest the targets were too easy; scores near 0 suggest they were unrealistic, or simply the wrong goals to chase.
Why is 0.7 the target for OKRs?+
OKRs are meant to be a stretch, so fully hitting 100% would mean you set them too safe. Aiming for roughly 0.7 keeps goals ambitious but achievable: you’re deliberately reaching for more than you’re sure you can do.
How do you score an objective with multiple key results?+
Average the key-result scores. An objective with three key results at 0.8, 0.6, and 0.4 scores 0.6. Read it as a health indicator of the objective overall, not a precise grade: it tells you “mostly there” versus “barely started.”
What’s the difference between confidence and score in OKRs?+
Confidence is a mid-cycle forecast (how likely you are to hit the key result) checked at each weekly review. The score is the end-of-cycle result, 0.0 to 1.0. Confidence tells you where to intervene during the quarter; the score tells you what to learn after it.
How do you score a key result that isn’t measurable?+
You mostly can’t, and that’s the signal. If a key result can’t be scored from a number, it was written as a task or an opinion, not an outcome. Rewrite it with a metric and a target before the next cycle so grading is arithmetic, not an argument.
What is an OKR scorecard?+
The whole cycle on one page: every objective, its key results, the baseline and target, where the number landed, the score, and one line on what you learned. It isn’t a separate method or a tool you have to buy: it’s the artefact you bring to the grading session so nobody is hunting for last quarter’s numbers mid-meeting.
How do you score a key result where the metric should go down?+
Measure the distance you closed, not the ratio. For churn from 8% to a 2% target that landed at 4%: (8 − 4) ÷ (8 − 2) = 0.67. Dividing the two numbers instead (8 ÷ 4) would read as 200% and quietly turn a partial win into a blowout.
What happens to the objective score if some key results were never scored?+
Divide by the number of key results the objective has, not the number that got scored: an unscored key result counts as 0.0. Averaging only the scored ones inflates the objective: one key result at 1.0 alongside two nobody touched reads as a perfect 1.0 when it is really a 0.33.
Is there a standard OKR scoring system?+
The common one is the 0.0–1.0 scale with roughly 0.7 as the target for a stretch, popularised by Google and used by most teams running OKRs. Some teams use percentages or a simple red/amber/green instead. Any of them works as long as the rule is written down before the cycle and applied the same way every time; what breaks a scoring system is changing it once the numbers are in.