What makes a KR a real KR
- Outcome, not activity. "Reduce time-to-value to under 5 minutes" not "ship onboarding redesign."
- Measurable. A binary or numeric value an outsider could verify.
- Time-bound. By the end of the quarter, this number = X.
- Limited count. 3–5 KRs per objective. More dilutes focus.
The activity-vs-outcome test
Ask: "Could we hit this KR and the company still be no better off?" If yes, it's a task disguised as a KR. "Ship the new dashboard" could be hit while no one uses the dashboard. "30% of weekly-active users open the new dashboard" can't be hit without success.
↳ the cascading-OKR antipattern
Some orgs cascade OKRs strictly: company OKRs → team OKRs → individual OKRs. This kills agility — by the time team OKRs are set, company OKRs are stale. Doerr suggests connect, don't cascade: team OKRs reference and ladder up to company OKRs but are set in parallel.
Quarterly cadence
Set at quarter start. Mid-quarter check-in to grade progress (typically 0.0–1.0). End-of-quarter scoring + retro on what worked and didn't. Quarter+1 OKRs reflect lessons. Annual OKRs at the company level can pair with quarterly ones at team level.