Writing OKRs

OKRs (Objectives and Key Results) are a tool for aligning outcomes. Used well, they focus teams on measurable change. Used poorly, they become a second roadmap of vanity metrics and sandbagged targets.

Objectives are qualitative, ambitious, time-bound aims. Key results are measurable evidence you are getting there — typically 2–4 per objective. Origin: Andy Grove at Intel; John Doerr spread OKRs to Google and beyond.

Good KRs are specific, measurable, owned and time-bound. Prefer rates and absolute counts over vague "improve." Baseline the current number in the KR text so progress is obvious: "from 22% to 35%," not "increase activation." Prefer leading indicators you can influence this quarter (activation, invite rate) alongside lagging ones (revenue), or you will steer by looking in the rear-view mirror.

Pair growth KRs with guardrail metrics named in the same OKR set. Avoid KRs teams can hit by gaming: "increase emails sent" without a quality or unsubscribe guardrail.

The big idea

Objective: New teams experience value in their first session.

KR1: Activation rate (created project + invited member within 24h)
     from 22% → 35% for new workspaces.
KR2: Median time-to-first-project from 48 min → 15 min.
KR3: Support tickets tagged "setup confusion" −30%.

Common anti-patterns:

• OKRs as a task list: "Ship checklist by May" is an output milestone, not a KR. Prefer the metric the checklist was meant to move.
• Too many: Three company objectives beat twelve. Focus is the point — see strategy.
• Sandbagging: 100% hit rate every quarter usually means targets were too soft. Stretch KRs that land ~60–70% are healthier in many cultures (Google's historical "grade").
• Cascading theater: Copying company OKRs verbatim to every team creates fake alignment. Teams need their problems that contribute to company outcomes.
• Conflicting KRs: Growth KR that rewards spammy notifications vs a guardrail on unsubscribe — make trade-offs explicit.
• Set and ignore: OKRs need mid-quarter check-ins. Dead OKRs teach cynicism.
• Individual OKRs that punish collaboration: Prefer team OKRs with individual responsibilities underneath.

Mid-quarter, schedule a 30-minute check: red/yellow/green per KR, one sentence on blockers, one decision. Do not wait until the retrospective to discover you never instrumented the metric. End-of-quarter, grade honestly and write one paragraph on what you will change in how you set KRs next time.

Walking through it

"Ship checklist by May" becomes the objective above. If the checklist ships and activation stays 22%, you missed the KR — which is the point. The team should then try templates or concierge, not declare victory because Jira is green.

For interviews: write one objective with three KRs and name one anti-pattern you avoided. Clarity beats buzzwords — say "outcome" and mean a number.

Common mistakes

• Task-list KRs. Shipping is output; the KR is the behavior change.
• Twelve objectives. Nothing is focused.
• 100% hit rate every quarter. You sandbagged.

Check yourself

1. Rewrite "Ship checklist by May" as a KR.

Activation rate (created project + invited member within 24h) from 22% → 35% for new workspaces — or another metric the checklist was meant to move.

2. Why label committed vs aspirational OKRs?

Without labels, a missed stretch looks like failure and a missed commit gets shrugged. They are different contracts.

Key takeaways

• Objectives are qualitative; KRs are measurable evidence with a baseline.
• Avoid task lists, too many objectives, sandbagging and fake cascades.
• Label committed vs aspirational; check mid-quarter; grade honestly.

Further learning

Measure What Matters — OKRsDo what matters with OKRs.What Matters

John Doerr on why the right goals matter (TED):

Next: fit OKRs to real capacity in roadmap capacity.