How to write good OKRs, with examples

Most OKRs fail before the quarter starts. Not because the team lacks ambition, but because what got written down was a task list with an inspiring title on top. This guide shows what a good OKR looks like, the rules that separate an objective from a key result, and how to rewrite the five most common mistakes.

#What is an OKR?

OKR stands for Objectives and Key Results. Andy Grove developed the method at Intel in the 1970s; John Doerr brought it to Google in 1999, and his book Measure What Matters made it the default goal-setting framework for tech companies.

The structure is simple. An objective says where you want to go. Two to four key results say how you will know you got there. Everything you actually do to move those numbers (projects, launches, experiments) sits underneath as initiatives.

#How to write a good objective

A good objective describes a state of the world you want to be true at the end of the period. It should be short enough to remember, specific enough to rule things out, and meaningful enough that the team cares whether it happens.

• Qualitative. Numbers belong in the key results. “Grow revenue 40%” is a key result wearing an objective’s clothes.
• Outcome, not activity. “Make onboarding effortless” rather than “Redesign the onboarding flow”.
• Scoped to what you can move. A VP can own “Become the default choice for mid-market teams”. An engineer on the onboarding squad cannot, but can own “Make the first hour in the product feel obvious”.
• Time-bound by the cycle. It should be achievable, or at least meaningfully advanced, within the quarter.

A quick test: read the objective aloud to someone outside your team. If they can’t tell what would be different at the end of the quarter, rewrite it.

#How to write key results

A key result is a measurable outcome with a metric, a baseline and a target. “Raise week-1 activation from 38% to 60%” has all three. “Improve activation” has none of them.

The rules that matter most:

1. Always state the baseline. A target without a starting point can’t be graded. If you don’t know today’s number, the first initiative is to measure it.
2. Measure the result, not the effort. Shipping a feature, holding training sessions or hiring a contractor are ways to move a number. They are not the number.
3. Keep units consistent. If the company key result is in USD of annual recurring revenue, team key results that roll up to it should be in USD of ARR as well.
4. Mix leading and lagging indicators. Revenue and churn (lagging) tell you what happened; pipeline, activation and cycle time (leading) tell you early whether you’re on track.
5. Two to four per objective. One key result is usually gameable; five or more means the objective is really two objectives.

#Key result or initiative? The one-question test

Ask: “If we did this perfectly and nothing changed for customers or the business, would we call it a success?” If the answer is yes, it’s an initiative. Turn it into the outcome it is supposed to move.

#OKR examples, before and after

Here are rewrites across three functions. The “before” versions are real patterns we see constantly; the “after” versions follow every rule above.

#Product and engineering

#Sales and marketing

#People and operations

#The five most common OKR mistakes

#1. Writing a to-do list

The most common failure. Every key result starts with a verb like launch, build, run or hire. The team will feel productive, grade themselves 100%, and nothing measurable will change. Fix it with the one-question test above.

#2. Too many OKRs

Seven objectives with five key results each is 35 things to track, which means none of them is a priority. OKRs are a tool for saying no. If everything the team does is in the OKRs, they’re a work log, not goals. One to four objectives per person is plenty; many strong teams pick one or two.

#3. Copying the CEO’s OKRs down the org

“Grow ARR to $10M (Marketing)”, “Grow ARR to $10M (Engineering)”. Copying a parent objective and appending a team name tells nobody what they own. Each child objective should state the part of the parent its owner can move. We cover this in depth in alignment vs cascading.

#4. No baseline, or a made-up target

“Increase NPS to 50” is meaningless if you don’t know whether today’s NPS is 10 or 48. Write the baseline next to every target. If a target is a guess, say so and revisit it at the first check-in.

#5. Set and forget

OKRs written in week one and opened again in week thirteen are a planning exercise, not a management system. A 15-minute weekly check-in, where each key result owner updates the number and flags risk, is what makes them work. Our week-by-week quarter playbook shows the whole cadence.

#Committed vs aspirational OKRs

Not every OKR should be a moonshot. It helps to label each one:

#A checklist for every OKR you write

• Is the objective qualitative and about an outcome?
• Can you name the single owner of each key result?
• Does every key result have a metric, a baseline and a target, in consistent units?
• Would doing every initiative perfectly, with no change in the number, count as failure?
• Is there at least one leading indicator you can see move within a few weeks?
• Does it connect to a parent OKR, or explain why it doesn’t?

#Writing OKRs with AI

Most of these rules are mechanical enough for software to check. OKR.ai’s AI coach drafts key results for any objective you write, refuses to turn activities into key results, and asks for baselines. When your manager has OKRs, it drafts yours to support theirs so the numbers add up. You edit everything before it’s saved. See how AI drafts OKRs from Jira, GitHub and your HR system, or try it on your own org in about five minutes.

#Frequently asked questions