How to Write OKRs That Teams Actually Commit To

Most OKR guides explain what an OKR is. This one explains how to write one. There's a grammar to it. Once you see the pattern, you spot the weak ones immediately. I'll walk through the structure, then rewrite three bad OKRs to good ones so you can see the move in practice. If you can write one solid OKR after reading this, I've done my job.

By Max Bondarenko · Last updated May 2026

The grammar of an Objective

An Objective describes the world you want to live in by quarter-end. It's qualitative. It should read like a sentence you'd say out loud.

[Verb] + [specific state you want to be true] + [optional: why it matters]

Example: "Make onboarding fast enough that customers hit value before the end of day 1."

The Objective is the destination. It does not include numbers. Numbers live in the KRs. The Objective gives the team a thing to point at. The KRs tell them whether they reached it.

Two tests I run on every Objective: would somebody on the team be genuinely energized reading this? And does it force a real choice about priorities, or is it generic enough to cover anything? "Improve product quality" fails both. "Make our API reliable enough that customers build production workflows on it" passes both.

The grammar of a Key Result

KR is a number with three parts: what you're measuring, where you are, where you want to be.

[Metric] from [baseline] to [target]

Example: "Day-7 retention from 31% to 55%."
Example: "Organic sessions from 18,000 to 42,000/month."
Example: "Pipeline coverage from 2.1x to 3.5x."

You can't skip the baseline. I've seen teams write "increase organic traffic" as a KR. That's not a KR. It's a direction. A KR has a number you started at and a number you're aiming for. The gap between them is what the team is actually doing this quarter.

Three rewrites

The vague one

Before

Objective: Improve customer experience.
KR1: Improve NPS.
KR2: Reduce support tickets.
KR3: Make onboarding better.

After

Objective: Make customers successful in their first week, before they have any reason to doubt us.
KR1: Onboarding completion 54% to 81%.
KR2: Day-7 retention 38% to 56%.
KR3: Post-onboarding NPS 22 to 44.

Why: Original is a wishlist. No baselines, no targets, "improve" three times. Rewrite has the same intent but you can tell on day 88 whether you hit it.

The task masquerading as a KR

Before

Objective: Scale our engineering team.
KR1: Hire 5 engineers.
KR2: Implement new interview process.
KR3: Set up engineering blog.

After

Objective: Build a team that can ship twice as fast in Q4 as we did in Q3.
KR1: Hire 5 engineers, all 4 ramped by Q3W12.
KR2: Code-merged to in-production: 3h 40min to 48min.
KR3: Engineering blog live, 3+ qualified applicants in Q3 mention it.

Why: Hiring 5 engineers is something you do. Whether the team is actually faster is the outcome. The rewrite measures the outcome.

The impossible one

Before

Objective: Dominate our category.
KR1: Become the #1 OKR tool.
KR2: Get 10x more customers.
KR3: Be the most talked-about product in our space.

After

Objective: Establish clear market leadership in AI-native OKR software.
KR1: G2 OKR category - Top 3 by Q3 end.
KR2: New customer signups 340 to 1,200 in Q3.
KR3: 5+ analyst or editorial best-of roundups mention us in Q3.

Why: First version is a mood. Rewrite has the same energy but you can prove it.

The five-question check before you ship it

Is the Objective inspiring out loud? Read it in an all-hands voice. Would the room feel something, or would they pick up their phones?
Does each KR have a baseline and a target? If not, you can't track it. Find the baseline before you commit to a target. I've watched teams skip this and end up arguing about whether they hit the KR or not.
Are the KRs outcomes, not outputs? "Ship X" is an output. "X% of users doing Y after we ship it" is an outcome. Big difference. The first one you can hit without anything actually working.
Is the target uncomfortable? 70% confidence is the right zone. If you think there's no chance you'll miss, you set it too low. If you think there's no chance you'll hit, you set it too high.
Would hitting all KRs mean you hit the Objective? If you can imagine a world where you green every KR and the Objective still feels unfinished, the KRs are measuring the wrong thing. Rewrite.

FAQ

How many OKRs should a company have per quarter?

3 to 5 at the company level. I have never seen a team have too few. I have seen plenty have too many. The first quarter I tried six, we landed about 60% on each because nothing got the focus it needed.

Should OKRs be tied to compensation?

I would not do it. The original Intel and Google design treats OKRs as a learning tool. The second you tie them to bonuses, every team sandbags their targets. Performance reviews exist for a reason. Use them for that.

How long should Key Results be?

One sentence. Under 15 words if you can. "Day-7 retention from 31% to 55%" is a good KR. If yours takes three sentences to explain, you have either two KRs hiding inside one or you are describing a project plan.

Can an Objective have only one Key Result?

Allowed but usually a smell. A single KR means you only measured success from one angle. Pick two more dimensions that would also have to be true for the objective to count.

How do you write OKRs for projects that span multiple quarters?

Slice the project into quarterly states. Each quarter describes the specific shape the project should be in by EoQ, not the final shipped product. "MVP with 5 pilot customers live" is a Q3 OKR. "Build the new product" is a roadmap line, not an OKR.

Write your OKRs and track them in Okiar — free during beta.

Okiar is free during beta. Voice check-ins, AI projections, team health — live in minutes.

Start free →