Definition (canonical)
OKR (Objectives and Key Results) is a goal-setting method pairing a qualitative, ambitious Objective with 2–4 measurable Key Results that define what "achieved" means. The format's power is the forced marriage: inspiration that can be graded — "Become the obvious choice for mid-market logistics" stays poetry until KRs like "win rate 25%→40% in shortlists" pin it to reality.
The mechanics that make it work
Cadence: quarterly OKRs against an annual frame — long enough to move something real, short enough to correct course. Structure: company OKRs set direction; teams write their own that CONTRIBUTE (not cascade mechanically — the alignment conversation is the value, a spreadsheet of decomposed numbers is bureaucracy wearing OKR's clothes). Focus: 1–3 objectives per team, hard maximum — OKRs are a prioritization technology; ten objectives means none. Grading: 0.0–1.0 at quarter end, with ~0.7 as the healthy landing for ambitious KRs — consistent 1.0s mean sandbagging, consistent 0.3s mean fantasy planning. The split that saves careers: committed OKRs (expected 1.0, consequences real) labeled separately from aspirational ones — mixing them teaches teams that all targets are negotiable.
OKRs vs KPIs — different instruments
KPIs monitor the running machine (dashboards, always-on, health); OKRs direct CHANGE (what we'll make different this quarter). They cooperate: a KPI going red often births an OKR ("churn 3%→1.5%"); an OKR achieved often retires into a KPI to hold the line. Companies fail by using one for both jobs — KPI-only stagnates on maintenance; OKR-only reinvents its priorities quarterly while the engine room goes unwatched. The north star sits above both as the unifying flag OKRs should ultimately move.
Failure modes — where OKR programs die
OKR-as-task-list: KRs like "launch feature X" measure output, not outcome — the feature can ship and change nothing; write KRs as state changes in the world. Set-and-forget: OKRs reviewed only at quarter end are decoration — weekly check-ins on confidence keep them steering. Performance-review fusion: grade OKRs for learning, pay people via separate mechanisms — bonus-linked OKRs get sandbagged within two quarters, guaranteed. Tooling worship: a shared doc beats an unadopted platform. And the meta-failure: adopting OKRs to fix a strategy vacuum — OKRs execute strategy; they cannot secrete one.
FAQ
How long until OKRs work? Two to three cycles of visible calibration errors — budget for awkward quarters; teams quitting after one messy quarter never reach the payoff.
Do OKRs fit a 5-person company? The conversation fits (what matters most this quarter?); the ceremony doesn't — one shared objective may be the whole system.
Individual OKRs? Mostly counterproductive below leadership level — team OKRs plus personal development goals separately keeps the tool honest.
Related: KPI · North Star Metric · Governance Reviewed: 2026-07-16 · Sources: OKR canon