// okr examples · nine functions

OKR examples, with the bad version shown too.

Every example here appears twice: the plausible version most teams write, and the one worth a quarter. The difference between them is the whole skill, and it is invisible in a list of good examples on their own.

Written by Matt Roberts, ZOKRI co-founder Last reviewed August 2026 Nine functions · nine anti-patterns
// how to read these

Copy the shape, never the content. A Key Result only means something against a baseline you actually have and a strategy you have actually chosen.

The numbers below are illustrative, drawn from the situations these goals came from. Yours will be different, and if you cannot state your baseline, that is the first thing the quarter is for.

And a card is not a goal record. The narrative, the reasoning under each number, the work, the obstacles and the retrospective are all missing here by design. What a full one carries is set out below the examples.

// Company
#

Mid-market SaaS, £14m ARR, growing but winning the wrong deals.

Become the obvious choice for finance teams leaving spreadsheets behind.

Win rate against spreadsheet-only incumbents 22% → 35%
Share of new revenue from the mid-market segment 31% → 50%
Days from first call to signed contract 74 → 45
The version we usually see

Objective: Grow revenue. KR: Hit £18m ARR. KR: Launch the new pricing page.

What changed, and why

Revenue is the score, not the play, so it belongs on the scorecard while the goal targets the drivers. The pricing page is an initiative. The new Objective also names where we are competing, which is the thing the old one quietly refused to decide.

// Product
#

B2B tool with strong trial numbers and weak conversion to real use.

New teams reach their first real outcome without needing us.

Workspaces activated within 14 days 38% → 60%
Support tickets per new workspace in month one 4.1 → 1.5
New workspaces completing setup unaided 44% → 70%
The version we usually see

Objective: Improve onboarding. KR: Ship the new onboarding flow by 30 June. KR: Increase NPS.

What changed, and why

Shipping the flow is how, not what: it can land in full while activation does not move. NPS is too slow and too broad to steer by inside a quarter. The rewritten Key Results describe the customer getting somewhere, which is the only thing the team is actually trying to cause.

// Engineering
#

Deploys are fine on Monday and nobody dares ship on Friday.

Shipping on a Friday stops being a decision anyone has to think about.

Change failure rate 14% → 5%
Median time to restore service 3h 40m → under 45m
Deploys per week 8 → 25
The version we usually see

Objective: Improve engineering quality. KR: Reduce bugs. KR: Migrate to the new CI system.

What changed, and why

Reduce bugs has no baseline, no target and no end. The migration is a piece of work, and it may or may not be the thing that fixes the problem, which is exactly the question the Key Results should leave open.

// Sales
#

Good pipeline, respectable close rate, and a pattern of losing deals the team expected to win.

Win the deals we should never have lost.

Win rate on qualified opportunities 24% → 33%
Deals slipping a quarter 19 → 8
Closed-lost with a recorded reason 40% → 95%
The version we usually see

Objective: Smash Q3. KR: £2.4m closed. KR: 200 discovery calls.

What changed, and why

Quota is a KPI and a commission plan, and putting it in the goal set adds nothing except pressure. Calls are activity. The third Key Result looks like admin and is not: without recorded loss reasons, next quarter's goal is guesswork.

// Marketing
#

Plenty of content, very little of it doing any work.

Buyers arrive already believing the thing we would otherwise have to convince them of.

New pipeline sourced from owned channels 18% → 35%
Demo requests citing a specific published piece 6 a month → 30 a month
Sales cycle on content-sourced deals 68 days → 50 days
The version we usually see

Objective: Increase brand awareness. KR: Publish 50 posts. KR: Reach 10,000 followers.

What changed, and why

Awareness as written cannot be proved or disproved. Post count is a workload and followers are vanity: nothing downstream changes when either moves. The third Key Result is the interesting one, because it tests whether the content is doing persuasion rather than traffic.

// Customer success
#

Renewals are holding, and the team is finding out about problems at renewal.

Customers reach the value they bought before anyone mentions renewal.

Accounts hitting their stated success criterion by day 90 41% → 70%
Gross revenue retention 87% → 93%
Accounts with a documented success criterion 30% → 100%
The version we usually see

Objective: Reduce churn. KR: Keep churn under 5%. KR: Run a QBR with every account.

What changed, and why

Churn under 5% is a threshold, which makes it a KPI. The QBR is a process commitment: worth doing, and not a result. Note that the third Key Result is close to a task and earns its place because nothing else on the list can be measured until it exists.

// People
#

Engagement scores are fine on average, and three teams are quietly haemorrhaging people.

Managers stop being the reason good people leave.

Regretted attrition in teams under first-time managers 19% → 10%
“My manager gives me feedback I can use” agreement 54% → 75%
Days from a flagged concern to a documented action 42 → 10
The version we usually see

Objective: Improve culture. KR: Run the engagement survey. KR: eNPS above 40.

What changed, and why

Running the survey is the instrument, not the goal, and a company average hides exactly the problem worth fixing. Segmenting the attrition measure to first-time managers is what turns a vague ambition into a bet someone can act on.

// Finance
#

The numbers are accurate and they arrive too late to decide anything with.

The leadership team can decide on Tuesday with numbers they trust.

Working days to close the month 12 → 5
Forecast variance against actuals 18% → 7%
Budget owners who can self-serve their position 20% → 80%
The version we usually see

Objective: Improve financial reporting. KR: Implement the new BI tool. KR: Produce a monthly board pack.

What changed, and why

The tool is an initiative and might not be needed. The board pack already happens, so committing to it is a Key Result the team could have graded green on the day it was written.

// Operations
#

The process works, as long as the same two people are in the building.

The core process runs the same way twice, without a hero.

Orders needing manual intervention 31% → 10%
On-time delivery 88% → 97%
Exceptions with a recorded root cause 15% → 90%
The version we usually see

Objective: Operational excellence. KR: Document all processes. KR: Zero errors.

What changed, and why

Operational excellence is a virtue, not a choice. Zero is a slogan rather than a target, and documenting is work. The rewritten set says what the documentation is supposed to achieve, which is the test the documents will eventually be judged against.

// what a card cannot show

An Objective and two or three Key Results is the tip of a goal record.

Everything above is the sentence. It is the part that fits on a card, the part software stores, and the part every examples page on the internet stops at. It is also the part that survives contact with reality least well, because none of the reasoning travels with it.

A goal we would actually install carries six layers. Writing them takes an extra hour at the start of a quarter and it is the hour that decides the other twelve weeks.

The choices behind the goal, written down. What we chose, what we deliberately ruled out, and the crux: the one pivotal, solvable problem this quarter is aimed at. A team that can read the narrative can make the hundred small decisions nobody will be in the room for.

“We are choosing mid-market finance teams and explicitly not enterprise this year. The crux is that we lose to spreadsheets, not to competitors.”

Two sentences under every Key Result: why this metric, and why this target. The belief that makes it the right measure, and the reasoning that makes the number credible, whether that is a baseline, a comparable, or a mechanism. Without it, next quarter nobody can tell a stretch from a guess.

“38% is our own last-90-days figure. 60% is where our two best-served segments already sit, so we believe it is reachable without a new product.”

Initiatives, Process Commitments and Experiments, held deliberately separate from the outcome. Scoped deliverables, recurring behaviours, and tests of a hypothesis. Do, experiment, repeat: the plan is a set of bets underneath a bet, and it is expected to change while the Key Result does not.

“Experiment: guided setup for the top three use cases, tested on 40 new workspaces, decision point in week five.”

Before the goal starts, the team imagines it has failed and says why. Every named cause becomes an entry in the issues and obstacles log with an owner and a date, kept alive in check-ins and drained by escalation. Confidence declared high against a long live obstacle list is a contradiction someone should have to explain.

“Named at creation: the data we need is owned by a team with no capacity until week six. Owner: Priya. Review: week two.”

A named OKR lead, supported rather than merely appointed, and a team drawn around the goal instead of the org chart. Where the goal needs it, the topology flexes for the quarter. And the honest part: what was stopped to make room, because a goal added to a full plate loses to the plate.

“Lead: Sam, with two days a week protected. Two engineers seconded from platform. The billing refactor is deferred to Q4 to fund it.”

Progress is the least interesting thing at the end. A real retrospective records three things: what the work delivered, what value actually landed, and what the team now believes that it did not believe twelve weeks ago. That last one is the asset, and it is the one almost nobody keeps.

“We now believe onboarding was never the constraint. Sales set the wrong expectation, and we have the evidence.”

Judge a goal by its record, not its sentence. Two teams can write the same Objective, and only one of them will be able to say, in twelve weeks, what they now know.

See the full method → Have someone in the room →
// nine that look fine and are not

Bad OKR examples, diagnosed.

These are all real shapes, all plausible in a planning session, and all gradeable green while the business stands still.

The Key Result Diagnosis What to do instead
“Launch the mobile app by 30 September” Task It can be completed in full while nothing about the business changes. Ask what the launch is meant to cause, then measure that.
“Maintain 99.9% uptime” Threshold Nothing is being changed, so there is no bet. It belongs on the KPI scorecard with an owner and an alert.
“Increase revenue” No baseline, no target Not measurable as written, and impossible to grade honestly. Every Key Result needs a from, a to and a date.
“Grow the following to 10,000” Vanity Pick any number that only goes up and nothing downstream depends on. It will be hit and nothing will follow.
“Improve customer satisfaction” Unfalsifiable Name the measure and the movement, or accept that the team will decide at the end whether it improved.
“Hire twelve engineers” Input dressed as outcome Hiring is how. The goal is whatever those engineers make possible, and twelve may be the wrong number for it.
“Complete the migration” Binary milestone Perfectly good as an initiative under a Key Result. As the Key Result itself, it puts the reward on finishing rather than on working.
“Achieve 70% of the target” Pre-sandbagged The 0.7 scoring convention leaking into the goal itself. Set the target you mean, label the stretch, and grade on evidence.
“Hold five customer interviews” Discovery as a goal Learning work belongs in its own continuous track. Grading learning is how you stop people learning.

The full catalogue of named failure modes, with the reasoning behind each, is in the glossary.

// the sentence patterns

Four templates, and the test that comes last.

The Objective

A sentence a team could repeat from memory, naming the change and who it is for, with no number in it.

“New teams reach their first real outcome without needing us.”

The Key Result

[Measure] from [baseline] to [target], by [date].

“Workspaces activated within 14 days, from 38% to 60%, by 30 September.”

When you have no baseline

Establish and publish the number this cycle, then set a real target for the next one. This is a baseline Key Result and it is a legitimate use of a quarter.

“Publish the first measurement of time-to-first-value by 15 August.”

The test before you commit

If it can be completed by doing work rather than by the world changing, it is an initiative and belongs underneath a Key Result.

“Ship the new flow” fails. “Activation from 38% to 60%” passes.

// asked and answered

Eight straight answers.

QWhat is a good OKR example?

A good example has an Objective anyone on the team could repeat from memory, and two to four Key Results written as a movement from a baseline to a target by a date. The test is whether the Key Results could all be achieved while the business stays the same. If they could, they are tasks. If they could not, you have a real goal.

QHow many Key Results should an Objective have?

Two to four. One usually means the Objective is really just that measure with a sentence on top. Five or more usually means the team has listed everything it plans to do rather than choosing what the goal is actually about. Three is the most common shape in practice.

QCan a Key Result be a project or a milestone?

Occasionally, and it is a warning sign in bulk. A milestone Key Result puts the reward on finishing rather than on the outcome moving, and it can be hit in full while nothing changes. Milestones are usually better placed underneath a Key Result as an initiative, where they describe the work rather than the result.

QShould we copy OKR examples from other companies?

Copy the shape, never the content. The measures on this page came from specific situations, and a goal only means something against a strategy and a baseline you actually have. The most common failure with examples is adopting a plausible-sounding Key Result nobody in the room can influence.

QWhat is an example of a bad OKR?

“Objective: improve onboarding. Key Result: ship the new onboarding flow by 30 June.” It is plausible, it is specific, it has a date, and it is a task. The team can deliver it perfectly and activation can stay exactly where it was, which means the goal never tested anything.

QHow do you write an OKR when you have no data?

Use a baseline Key Result: commit to establishing and publishing the measurement this cycle, then set a real from-X-to-Y target next cycle. It is an honest use of a quarter and far better than inventing a target nobody believes, which teaches everyone that the numbers in goals are decorative.

QWhat is the difference between a company OKR and a team OKR?

A company OKR names the change the whole business is making and is usually owned by the leadership team. A team OKR is that team's own bet, written against the strategy rather than sliced out of the company's numbers. Teams should write their own, which is alignment rather than cascade.

QShould individuals have OKRs?

Personal goals are legitimate and belong in their own lane, not as a third tier of the company goal document. When individual OKRs are just the team's Key Results divided by headcount, they add administration and remove ownership. Development goals for the person are a different and better use of the format.

QWhat should a complete OKR record contain?

More than an Objective and its Key Results. A goal record we would install carries six layers: the Objective narrative with the choices made and the crux it targets; context under every Key Result saying why that metric and why that target; the work layer of initiatives, process commitments and experiments; a pre-mortem whose output seeds an owned, dated obstacles log; the resourcing, including the named lead and what was stopped to make room; and a retrospective that records value and learning, not just progress.

// go deeper
OKR Glossary →
101 terms, plain English
KPI vs OKR →
Which one is this metric?
Task-based Key Results →
The most common mistake
OKR Coaching →
Someone in the room when you write yours
// test your own

You have read nine good ones. The useful question is what yours look like.

The free AI OKR Coach checks your actual goals against the standards on this page, and rewrites the ones that fail.

Try the free coach → Meet us instead
// that worked

That Worked

Short reports from the gap between plausible and works.

Examples are the free layer. What we're learning about using them is the fortnightly one.

Issue #7: Calibrated to be hit. Read previous issues →

Fortnightly. Under 500 words. Every claim carries a receipt. Unsubscribe is one click and no hard feelings. Privacy.

Ready to put the words to work?

Start with a look at your goals. A couple of weeks, small fixed fee, honest answers.

Meet Us
// the install pack

You know the words. Now test the sentences.

The Coach, our free install, checks whether your actual OKRs live up to these definitions, and rewrites them when they don’t.

Get the book and pack →
// the book and the install pack

Six things turn a company that trades into one that scales.

An hour to read, five working templates and the AI install pack. Ten years of strategy, operating systems, goal frameworks and culture building, battle tested in engagements that increased growth through better use of people, time and opportunity.

Get it for £100 →The only place the install pack is available.