OKRs, KPIs and performance reviews: making them work together

Every HR leader eventually faces the question: should OKRs feed performance reviews? Purists say never — the moment a stretch goal affects compensation, people stop stretching. Pragmatists ask what reviews are measuring if not progress against goals. Both are pointing at something real, and the practical answer is more useful than either slogan: connect them deliberately, with the relationship visible to everyone.

Three instruments, three jobs

OKRs are for direction and ambition: a small number of objectives with measurable key results, set at a cadence, expected to be missed sometimes — a 70% hit rate on genuinely ambitious key results is healthy. KPIs are for the steady state: the metrics a role must keep in range regardless of this quarter's objectives — response times, quality rates, pipeline coverage. Reviews are for the person: a periodic, evidence-based judgment of performance in role, which includes but is not limited to goal outcomes.

Instrument	Job	Cadence	Healthy result
OKRs	Direction and ambition	Quarterly or half-yearly	~70% attainment on stretch key results
KPIs	Steady-state performance of the role	Continuous	Metrics held in range
Reviews	Evidence-based judgment of the person	Per cycle	Scores people recognise as fair

The failure mode of full separation

Keeping goals entirely out of reviews sounds clean until you watch it operate. Reviews detached from goals drift toward impressions — how visible was this person, how smooth were they in meetings, what does the manager remember from the last six weeks? That is exactly the recency and likability bias a goal framework exists to counter. And employees notice fast when the OKRs they sweated over all year are absent from the conversation that decides their raise; the goals stop mattering within two cycles.

The failure mode of full coupling

Scoring reviews mechanically from key-result attainment is worse. Sandbagging becomes rational: the employee who commits to 80 and delivers 85 outscores the one who commits to 150 and lands 120. Context disappears — the market shifted, the dependency slipped, the target was inherited. A formula cannot see any of that. Attainment percentages are an input to judgment, never a substitute for it.

A connection that works

Make goal outcomes one weighted criterion among several in the review — visible, but not the whole score. Review goal difficulty at setting time, in the open: a manager who signs off on a goal as "ambitious" in January cannot call the 80% result a failure in December. Score the how alongside the what, using criteria like collaboration and craft that catch the goal-hitter who burns every bridge. And let the review narrative explain variance: the review is precisely the place where a human accounts for what the numbers cannot.

The connective tissue is data

None of this works if goals live in one tool, KPIs in dashboards and reviews in documents, stitched together by managers copy-pasting under deadline. The stitching is where context dies. In MeritFlo, goals cascade from organisation to individual, progress accumulates through the year, and when the review cycle opens, each goal arrives in the review with its history attached — commitment, difficulty flag, trajectory, outcome. The reviewer starts from evidence and spends their time on judgment, which is the one thing only they can add.