How to Write OKR Key Results That Measure Outcomes

TL;DR. A good key result describes a measurable change in an outcome. It does not describe the work you plan to do. To write one, name the metric that shows progress toward the objective, record where it stands today (the baseline), set the level you aim to reach (the target), and give it one owner, a defined data source and a review date. If you could finish the work and still miss the key result, you are measuring an outcome. If finishing the work guarantees success, you are measuring an activity.

What makes a key result different from an activity?

An objective says where you want to go. Key results show whether you got there. Activities, sometimes called initiatives, are the bets you place to move the key results.

The trouble starts when a team lists its plan and calls it a set of key results. "Launch the new onboarding flow" is a task. It can be completed on schedule while customers still fail to reach value. "Increase the share of new accounts that complete setup within their first week" is an outcome. It tells you whether the onboarding work actually helped.

A quick way to sort the two: ask "so what?" about each line. If the answer points to a change in customer or business behaviour, follow that answer until you reach something you can measure. That measurement is your key result. The original line belongs in the initiatives list.

If you are still deciding which numbers belong in OKRs and which are ongoing health metrics, read OKRs vs KPIs first. This post assumes you already know what you want to change and need to express it well.

What five elements does every key result need?

A key result that only names a metric is half-written. Before it goes into the plan, fill in all five of these elements.

Element	What it answers	Why it matters
Baseline	Where does the metric stand today?	Without a starting point, nobody can judge whether the target is ambitious or trivial
Target	What level are we aiming for by the end of the cycle?	Defines success clearly enough that people cannot argue about it afterwards
Owner	Who is accountable for tracking and reporting it?	Shared ownership often turns into no ownership. One name, even when many people contribute
Data source	Where does the number come from, and how is it calculated?	Prevents mid-cycle disputes about definitions and stops people quietly changing the formula
Review date	When will we check progress, and when is the final read?	Keeps the key result visible in the weeks between planning and grading

If you have no baseline, do not guess. Make "establish the baseline" an early task. Then set the target once you have a first reading, and record that the target was set late. Inventing a starting number makes the target meaningless.

Write the data source precisely. "CRM" is not enough. "Opportunities created in the CRM with source set to partner referral, counted on the date created" is. Precise definitions are what let different people produce the same number.

How do you turn weak key results into strong ones?

The examples below are illustrative B2B cases. The numbers are invented to show the structure and are not benchmarks.

Weak key result	Why it is weak	Stronger key result (illustrative)
Launch the new onboarding flow	Activity; done once it ships, whatever the effect	Raise the share of new accounts completing setup within 7 days from 40% to 55%. Owner: onboarding PM. Source: product analytics setup event. Reviewed every two weeks
Publish 12 blog posts	Counts output, not effect on pipeline	Increase sales-accepted opportunities that first engaged with content from 15 to 25 per quarter. Owner: content lead. Source: CRM first-touch field
Improve customer satisfaction	No metric, baseline or target	Reduce support tickets per active account from 3.0 to 2.2 per month. Owner: support manager. Source: helpdesk report, active accounts from billing
Run 20 customer interviews	Activity; interviews are an input to learning	Validate or reject three pricing hypotheses, each with a documented decision, before the pricing review. Owner: product lead. Source: decision log
Grow revenue	Too broad for one team to influence or own	Increase expansion revenue from existing accounts from a monthly baseline of 20k to 30k. Owner: account management lead. Source: billing system expansion line items
Reduce churn by working closely with at-risk accounts	Mixes method and outcome; no number	Lower monthly logo churn among mid-market accounts from 2.5% to 1.8%. Owner: customer success lead. Source: billing cancellations ÷ start-of-month active accounts

Notice that the stronger versions leave the method open. The onboarding team might redesign the flow, add guided setup or change default settings. The key result does not care which, so the team can switch tactics if the first bet does not work.

For churn and retention definitions, see Logo Churn vs Revenue Churn and the SaaS Metrics Glossary. For more marketing-specific sets, see OKR Examples for B2B Marketing.

How can you test whether a key result is good?

Run each draft through these checks before you commit to it.

• Outcome test: Could the team complete every planned task and still miss it? If not, it is an activity.
• Number test: Does it have a baseline and a target someone outside the team can check?
• Influence test: Can the owning team meaningfully move it within the cycle, even if not fully control it?
• Definition test: Would two people working from the stated data source get the same number?
• Objective test: If this moves, would you believe the objective is closer? If not, it measures the wrong thing.
• Gaming test: Could someone hit the target in a way that hurts the business, such as lowering quality to raise volume? If so, add a counterbalancing key result.
• Timing test: Will the data be available before the cycle ends? Metrics with long delays may need a leading indicator instead.

Three to five key results per objective is a common working range. Fewer keeps attention focused. If you need more, the objective may be too broad.

How should you handle outcomes that lag behind the work?

Some outcomes move slowly. Renewal rates, for example, may not respond within one cycle. In that case, choose a leading indicator you have reason to believe predicts the lagging one, and say so explicitly.

For example, if renewals depend on regular product use, a key result on weekly active usage among accounts approaching renewal is a reasonable proxy. Record the assumed link. At the end of the cycle, check whether the leading indicator and the lagging outcome actually moved together. If they did not, change the proxy next cycle.

What are the common mistakes when writing key results?

Listing tasks. "Ship", "launch", "run", "publish" and "hold" usually signal activities. Keep them in the initiatives list.

Skipping the baseline. A target of "reach 200 qualified leads" means little if nobody knows whether you are at 50 or 190.

Using vague verbs without numbers. "Improve", "optimize" and "strengthen" are directions, not results.

Choosing metrics the team cannot influence. Company-wide revenue as a key result for a single product squad leads to frustration. Pick the part of the outcome the team can move, or use a shared cross-functional OKR where several teams genuinely share it.

Changing the definition mid-cycle. If the formula changes, document the change and restate the baseline. Do not let the number quietly improve because the rules moved.

Setting and forgetting. A key result nobody reviews until the last week is just a wish. Agree a check-in rhythm at planning time. OKR Confidence Check-ins covers how to run those reviews.

Treating a miss as failure. Ambitious targets are sometimes missed. What matters is learning why. Run a Missed OKR Retrospective explains how.

When does an activity-based key result make sense?

Occasionally a milestone is the right measure. Examples include a regulatory requirement, a contractual delivery date, or work that has to happen before anything can be measured, such as setting up tracking. Use these sparingly, label them clearly as milestones, and pair them with an outcome key result wherever possible. If most of your key results are milestones, you have a project plan rather than OKRs.

What else should you know about writing key results?

Frequently asked questions

How do you write OKR key results?

+

Name a measurable outcome that shows progress toward the objective, then record its baseline, target, owner, data source and review date. Keep the work itself in a separate initiatives list.

How many key results should each objective have?

+

Three to five is a common working range. Fewer keeps the team focused. If you need many more, consider splitting the objective.

What if we have no baseline data?

+

Make establishing the baseline an early task, take a first reading, then set the target and note that it was set during the cycle. Do not invent a starting number.

Can a key result be a yes-or-no milestone?

+

Sometimes, for required deliverables or groundwork such as instrumentation. Use them rarely and pair them with an outcome measure where possible.

Who should own a key result?

+

One named person who tracks and reports it, even when several people contribute to moving it. Single ownership keeps follow-up clear.

If you want help applying this to your own team, talk to us.

Related reading

• OKRs vs KPIs — separate goals for change from ongoing health metrics.
• OKR Confidence Check-ins — review progress on key results during the cycle.
• OKRs for SaaS Teams — apply OKRs across product, growth and customer teams.
• OKR Examples for B2B Marketing — see complete objective and key result sets for marketing.
• Shared Cross-Functional OKRs — handle outcomes that several teams move together.
• Run a Missed OKR Retrospective — learn from key results you did not hit.