Set 2 to 3 company-level objectives with 3 to 5 measurable key results each, on a quarterly cycle. Use two layers (company and team) and skip individual OKRs for the first few cycles.
The structure
- Company objectives: two to three, qualitative and ambitious
- Key results per objective: three to five, quantitative and time-bound
- Layers: company and team only; team leads translate company objectives into their own
- Cadence: quarterly, with weekly 15 to 20-minute check-ins
- Scoring: 0.0 to 1.0, aiming for 0.6 to 0.7 average
- OKR champion: one named owner who runs check-ins and holds teams accountable
Your first quarter
- Weeks 1 and 2: leadership drafts two to three company objectives and shares the reasoning. Each team then proposes three to five measurable key results, like What Matters suggests. Push for "reduce response time from 24 hours to 4 hours" over "improve customer service."
- Weeks 3 to 12: publish OKRs visibly, hold weekly 15-minute stand-ups that score progress and surface blockers, and run a more in-depth monthly review.
- Week 13: score each key result on the 0.0 to 1.0 scale, discuss what worked, and draft the next quarter's OKRs.
Common mistakes to avoid
- Too many OKRs: More than three company objectives splits focus
- Vague key results: "improve satisfaction" is not measurable but "increase NPS from 32 to 45" is
- Using OKRs as a to-do list, since OKRs define outcomes
- Skipping check-ins
- Treating 70% completion as failure: 0.6 to 0.7 is the target
- Lacking an OKR champion. Mooncamp reports that companies typically have an OKR coach to run the process.
OKR implementations more often fail from overcomplication than from lack of ambition, and fewer, better OKRs win. Companies that need to set their OKRs typically evaluate the conceptual contrast with metrics that monitor health rather than drive change, like OKRs vs KPIs.
TL;DR
- For 20 people: two to three company objectives, three to five key results each, quarterly cycles, two layers, weekly 15-minute check-ins.
- One OKR champion owns the process. Score 0.0 to 1.0, aiming for 0.6 to 0.7.
- OKR implementations more often fail from overcomplication than from lack of ambition, and fewer, better OKRs win.