OKRs

OKRs separate the qualitative ambition (the objective) from the quantitative scoreboard (the key results). The discipline is harder than it looks: most companies that say they use OKRs are doing something else and calling it that.

Origin

OKRs were developed by Andy Grove at Intel in the 1970s, codified in his 1983 book High Output Management, and brought to Google by John Doerr in 1999 — Google has used them every quarter since. Doerr's 2018 book Measure What Matters brought OKRs into broader business consciousness. Grove's original framing emphasized two questions: "Where do I want to go?" (objective) and "How will I pace myself to see if I'm getting there?" (key results).

What an OKR actually is

An objective is a qualitative, ambitious, time-bound description of what you want to accomplish. "Make our SMB onboarding world-class." "Establish a defensible position in the legal vertical." "Become the default tool for video producers."

Key results are 2-5 numeric measures that, taken together, prove the objective was met. "Increase activation rate from 38% to 55%." "Reach 500 paying customers in the legal vertical." "Achieve 60% week-2 retention for new accounts."

The discipline is in the conjunction: if you hit all your key results but didn't accomplish the objective, your KRs were wrong; if you accomplished the objective without hitting the KRs, your objective was poorly written.

When OKRs are the right tool

OKRs work best in: companies of 30+ people that need cross-team alignment without micromanagement; environments with strategic ambiguity where direction-setting matters; cultures where missing a goal isn't punished but is examined; and businesses where outcomes are reasonably measurable on a 90-day cadence.

OKRs are a poor fit for: very small teams (the overhead exceeds the alignment value); operationally repetitive work (use SLAs and metrics, not OKRs); cultures that punish missed goals (people will sandbag); and work where outcomes are inherently slow (multi-year R&D doesn't fit a quarterly rhythm).

How to write good OKRs

1. Limit ambition. No more than 3 objectives per team per quarter; no more than 2-5 KRs per objective. The point is focus, not coverage.
2. KRs measure outcomes, not activity. "Launch the new pricing page" is an activity. "Increase conversion rate on pricing page from 4.2% to 6.5%" is an outcome. Activities go in your project tracker, not your OKRs.
3. KRs must be quantitative. If a KR can't be measured, it's an objective. "Improve customer satisfaction" is an objective; "Increase NPS from 32 to 45" is a KR.
4. Set ambitious targets. Grove advocated 70% achievement as a healthy hit rate — 100% on every KR every quarter means targets are too soft. Google publicly says "we aim for 0.6-0.7 score; 1.0 means we sandbagged."
5. Cascade with care. Top-level OKRs should inform team OKRs but not mechanically decompose into them. Each team needs autonomy to choose how to contribute, or OKRs become assigned tasks.
6. Score and review honestly. Quarterly grading (typically 0.0-1.0 per KR, averaged per objective) only works if grades are taken seriously without being punitive.

Worked example: a Series B SaaS company

A 90-person Series B company has identified that net revenue retention is the limiting factor on growth efficiency.

Company objective: Establish a self-sustaining growth engine inside our existing customer base.

• KR1: Increase Net Revenue Retention from 108% to 125%.
• KR2: Reduce annual logo churn from 14% to 8%.
• KR3: Achieve 35% of new ARR from existing-customer expansion (up from 19%).

The Customer Success team's quarterly OKR cascades from this:

CS objective: Make every Series B+ customer measurably more successful within 90 days of onboarding.

• KR1: 80% of new accounts complete the integration milestone within 30 days (current: 52%).
• KR2: Quarterly business review completion rate at 90% for accounts >$50K ARR (current: 64%).
• KR3: Win-back rate on "at-risk" accounts at 60% (current: 38%).

The Product team's OKR addresses the same company-level outcome from a different angle:

Product objective: Make expansion the path of least resistance for engaged customers.

• KR1: Self-serve seat expansion live and used by 30% of customers with >5 active users.
• KR2: Launch the analytics module to 80% of eligible accounts; achieve 25% activation.
• KR3: In-app usage signals trigger CSM outreach for 90% of expansion-ready accounts.

Two teams, three OKRs each, all visibly contributing to the same company objective without being assigned identical work.

How OKRs go wrong

• Too many. Companies that set 8 objectives with 5 KRs each are setting 40 priorities. There are no priorities; this is a to-do list.
• Activity dressed up as outcome. "Ship feature X" disguised as "Increase feature X adoption to 50%" — when adoption is determined by whether feature X exists, the KR collapses into the activity.
• Performance review entanglement. Tying compensation to OKR achievement guarantees sandbagging. Use OKRs for direction and learning; use separate processes for individual evaluation.
• Cascading without translation. Mechanically pushing top-level KRs to teams produces metric stuffing. Each level should ask "how do we contribute to that?" not "what fraction of that number is mine?"
• No mid-quarter check-ins. OKRs reviewed only at quarter end produce surprise misses. Weekly or bi-weekly grading inside the team makes them useful as a leading indicator.

Critique

OKRs assume the world is measurable on quarterly cadences and that ambitious goals motivate rather than demoralize. Both assumptions hold less well in practice than in theory. They also bias toward measurable outcomes, which can crowd out non-quantifiable strategic work (research, exploration, relationship-building). Used as a focus tool they're excellent; used as a complete planning system they create blind spots.