okr-planning

OKR Planning (objectives, key results, and the review cycle)

Three distinct objects — most bad OKRs confuse two of them

Objective  = qualitative, directional, memorable. Where are we trying to get to?
Key Result = measurable evidence the objective happened. metric: baseline → target
Initiative = the work you will do to move a KR. A project. Never a KR.

Test every line: if it can be marked "done", it is an initiative; if it can only be measured, it may be a KR. "Ship the referral programme" is an initiative; "referral-sourced signups from 4% → 12% of new signups" is the KR it serves. Objectives without KRs are slogans; KRs with no initiative are wishes.

Outcome, not output

A KR should describe a change outside your team — user behaviour, revenue, quality, retention, cost — not the volume of work the team emitted. Ten shipped features are output; the retention or conversion change they cause is the outcome. Output KRs are attractive because they are controllable, which is why they never falsify the strategy: you can hit all of them and learn nothing.

Writing a measurable KR

Each KR needs the metric, its exact definition and data source, a baseline measured before the quarter, a target, and one named owner. Missing baselines are the most common defect — without one, nobody can say at review whether the target was ambitious, trivial or already met. Prefer metrics already in a trusted dashboard; a KR whose measurement must first be built will not be measured, and rates beat raw counts, which drift with volume.

Keep to roughly 3–5 KRs per objective and few objectives per team; the limit is the point, because OKRs are a prioritisation instrument and a list of twelve is a list of none. Routine matters — uptime, support SLA, compliance — are health metrics you watch, not OKRs you push.

Lagging indicators (revenue, retention, NPS) confirm the result but move too late to steer inside a quarter. Leading indicators (activation rate, time-to-first-value, pipeline created) move early and are actionable but are only proxies — a proxy optimised in isolation is how a team hits its leading KR while the lagging one worsens. Pair them: one lagging KR to prove the objective, one leading KR actionable next week, and the causal assumption between them written down so the review can test it.