TL;DR: OKRs and performance reviews are two tools with two different jobs. OKRs make ambitious work visible while it is still in flight, and reviews assess a person over a longer arc. Connect them and people start setting goals they know they can hit, because an honest 0.6 now carries a personal cost. You keep accountability by keeping scores off the review form and making the weekly check-in the place where performance gets seen.
In this post
- Should OKRs be tied to performance reviews?
- The question comes up in nearly every rollout
- Why leaders connect OKRs and performance reviews in the first place
- What changes once the link is made
- What you give up
- How to separate OKRs and performance reviews without losing accountability
- If you are already in a gamed cycle
- Frequently asked questions
- Where to go next
Should OKRs be tied to performance reviews?
No. Keep OKR scores out of the formal review and out of compensation decisions. Once a score affects someone’s rating or their pay, they will set targets they are confident of hitting, and the early warning signal that made OKRs useful goes with it. Accountability still belongs in the system. Put it in the weekly check-in, where a stalled key result gets discussed while there is still time to act on it. A scorecard reviewed in month four arrives too late to help anyone.
The question comes up in nearly every rollout
Somewhere around week six of a rollout, a version of this question lands. In the rollouts I get brought into, it comes from a senior leader (the CFO or the head of HR, more than anyone), and it arrives in good faith.
If these are the priorities we said matter most, why aren’t they in the review?
It is a fair question. I keep watching leadership teams arrive at it independently, and the logic holds together right up until it meets a real team. You have spent a quarter telling everyone that OKRs represent the work that matters most. Then review season arrives and the form asks about something else entirely. The disconnect is genuinely awkward. Connecting OKRs and performance reviews looks like the obvious repair.
I have a lot of sympathy for the instinct. It comes from wanting the system to be coherent, which is the right thing to want.
The trouble is what the repair does to the numbers underneath it.
Why leaders connect OKRs and performance reviews in the first place
Three reasons come up over and over, and none of them are unreasonable.
The coherence argument. Two systems that both claim to measure what matters, running in parallel, never referencing each other. That is a strange thing to defend in an all-hands.
The accountability argument. Without a consequence attached, what stops a team from ignoring their OKRs entirely? Plenty of leaders have watched a program fade after one quarter and concluded the missing ingredient was teeth.
The fairness argument. Someone who delivered a hard objective ends up with the same rating as someone who delivered nothing. That feels wrong, because it is wrong.
Each argument is pointing at something real. Linking OKRs and performance reviews solves none of them, though, and it takes something valuable away in exchange.
What changes once the link is made
Teams are perceptive. Once OKRs and performance reviews share a form, people work out the new rules faster than leadership expects, within a cycle or two in the programs I’ve seen.
The first thing that shifts is target setting. A 0.7 that reads as a stretch goal honestly pursued invites people to set stretch goals. Put that same 0.7 on a form that shapes a raise and it reads as a missed target, so people set targets they know they will hit. The ambition drains out of the goal-setting session without anyone deciding to be dishonest. It is a rational response to a changed incentive.
The second shift is in what gets reported. A key result that is genuinely at risk in week five is the single most useful piece of information in an OKR program. It arrives while there is still time to move resources, cut scope, or ask for help. Once scores carry personal weight, that same signal becomes something to manage. The at-risk KR gets reframed, or reforecast, or explained. By the time it surfaces honestly, the quarter is mostly spent.
The third shift is the one leaders notice last. Scores start clustering at 1.0. From a distance, a cycle where nearly everything lands at full completion looks like a healthy program. Up close, the targets turn out to have been set low enough to guarantee it.
Goal-setting research spotted this pattern long before OKRs existed. Studies of aggressive goal setting have flagged the same side effects for years: narrowed focus, distorted risk-taking, and a measurable rise in unethical reporting when targets carry high personal stakes. OKRs are a goal-setting system, so they inherit its failure modes once you attach stakes to the number.
Performance reviews also carry problems of their own. Gallup’s research found that only 14% of employees strongly agree their performance reviews inspire them to improve, and that badly run reviews make performance worse about a third of the time. Feeding OKR data into that process makes the OKR data worse and leaves the review about where it started.
What you give up
| What you had | What the link costs you |
|---|---|
| Honest mid-cycle signal | At-risk KRs get managed out of sight |
| Ambitious targets | Targets set at the level the team is sure to hit |
| A useful score distribution | Scores bunched at 1.0 with nothing to learn from |
| Fast course correction | Problems surface in the retro, weeks after they started |
| A tool the team owns | A form the team completes for someone else |
The last row is the one that ends programs. A team that helped build its OKR treats it as a commitment. An OKR set to protect a rating turns into paperwork, and people can tell the difference immediately.
An OKR set to protect a rating turns into paperwork, and people can tell the difference immediately.
How to separate OKRs and performance reviews without losing accountability
Accountability survives the separation. It moves into the weekly rhythm, where it does more good.
Score the work, assess the person, and keep the two conversations apart. A key result scores 0.6 because the number moved from 40% to 58% against a 70% target. That is a fact about the work. Whether the person who owned it performed well is a separate judgment that includes how they responded when it stalled, who they pulled in, and what they learned. Leaders can and should form that judgment. It just should not be read off the score.
Put the accountability in the weekly check-in. The real pressure belongs here. Every key result answers one question: from what, to what? Take the metric your KR tracks and give last week’s value next to this week’s. A good answer sounds like “Activation went from 34% to 38%,” and the story about the new onboarding flow comes after the number. Answering that question out loud every week makes a team more accountable than any annual form can, because there is nowhere to hide for eleven weeks. Our OKR check-in template walks through the format.
Say the separation out loud, more than once. Teams that have been burned before will assume OKRs and performance reviews are still connected, whatever the policy says. It takes a few cycles of visible non-use before people believe it. Name it at kickoff, name it again when scores come in, and say exactly what the scores are used for.
Use the score distribution to diagnose the program. When everything lands between 0.9 and 1.0, that tells you how targets are being set, and it says very little about how hard any one person worked. Scores below 0.4 across the board point at capacity, or at objectives that were never connected to anything real. Either pattern deserves a conversation with the team, and neither one belongs in anyone’s rating.
Fix the coherence problem at the objective level. The original complaint was fair: it is strange for the review to ignore the work the company said mattered most. So connect each OKR to a real anchor from the start, and the work will show up in the review as work, with the score left out of it. The Link pillar of the LEAD framework exists for exactly this.
None of these five moves rescues a program on its own. Keeping scores out of reviews removes an obstacle. A working OKR program still runs on a weekly cadence, goals co-written with the team and leaders who keep paying attention, and this decision builds none of that. Treat it as one decision inside a much longer discipline. Make it anyway, because the alternative corrupts every number the program produces, one target-setting session at a time.
If you are already in a gamed cycle
Plenty of organizations linked OKRs and performance reviews years ago and are reading this from the other side, with two years of 0.95 averages and a leadership team that half-suspects the numbers are fiction.
The recovery is slower than the break, and it runs on proof. Announcing the new rule is the easy part.
- Announce the change before the next planning cycle starts. People need to set the next round of targets under the new rules for the change to mean anything. Waiting until scores land is too late.
- Do not reopen last cycle’s scores. Re-litigating old numbers teaches everyone that scores are still live currency.
- Expect the first honest cycle to look worse. Averages will drop. That is the system starting to tell the truth, and leadership needs to say so before someone reads the dip as a performance decline.
- Have a leader model it first. An executive presenting an honest 0.5 on their own objective, without flinching, does more than any policy memo.
- Give it three cycles. Trust in a scoring system rebuilds at the speed of demonstrated consistency.
If the scoring mechanics themselves need rebuilding, we go deeper on that in how to run an OKR scoring system without gaming it.
Frequently asked questions
Can OKRs and performance reviews ever reference each other?
Loosely, yes. A manager can reasonably discuss how someone handled a difficult objective, what they did when a key result stalled, and how they worked across teams. That is a conversation about behaviour and judgment. Importing the numeric score into the rating is the part that causes the damage.
What about team bonuses tied to company OKRs?
Company-level OKRs tied to a broad profit-share or company bonus are lower risk, because no individual can protect their own number by lowering a target. The damage scales with how directly a person controls the metric that pays them. Team and individual OKRs attached to individual comp are where sandbagging shows up fastest.
If scores do not affect anything, what stops people from ignoring OKRs?
Visibility and cadence. A key result discussed out loud every week by the people who own it, in front of the people who depend on it, carries more weight than a score reviewed twice a year. Adoption comes from OKRs being useful in the work every week, and that holds up far better than a consequence bolted onto the end of the year.
What score should a healthy team be averaging?
Somewhere in the 0.6 to 0.8 range across a set of ambitious objectives, with real variance between them. A team averaging 1.0 is almost always setting targets it has already hit. At 0.3, the team is either badly over-committed or working on objectives disconnected from what the business needs.
Does this apply to smaller companies too?
It applies more, if anything. In a 30 person company the founder can see who is delivering without a scorecard, so the score adds nothing to the review, and the incentive to sandbag lands just as hard.
Where to go next
If you want a read on which part of your OKR practice is costing you the most, the LEAD Toolkit walks through all four pillars and includes the self-assessment we use to find the weakest one. Some teams that arrive worried about scoring find the real gap somewhere else entirely.
Separating OKRs and performance reviews is a small structural decision that takes one conversation at kickoff. It protects the honesty of every number your program produces for the rest of the year. If Q1 planning is already on your calendar, that kickoff is a good place to say it out loud.
