OKR prompts for ChatGPT and Claude

These 14 OKR prompts help you write, check and run objectives and key results with ChatGPT or Claude: drafting objectives from strategy, turning them into measurable key results, aligning team OKRs with company goals, running weekly check-ins and scoring the quarter. Each prompt asks for your real numbers and context, because OKRs written without them measure nothing.

How to use these prompts

Every prompt works in ChatGPT, Claude or Gemini. Replace the blanks in double braces, or keep the default after the colon: {{period:Q1 2027}} means "Q1 2027" unless you change it.

Gather three things first: your company or team strategy for the period, last period's results, and the metrics you can actually measure today with a baseline. The prompts below ask for them.

The definitions these prompts use come from What Matters, the OKR site founded by John Doerr: "An Objective is simply what is to be achieved, no more and no less," and "Key Results benchmark and monitor how we get to the Objective." The same page says OKRs are typically written with one objective and 3 to 5 key results.

OpenAI's prompt engineering guide for everyday work suggests telling the model what "good" looks like and how you will decide it is correct. For OKRs, that means stating your rules (outcomes not tasks, a baseline for every number) in the prompt.

Which prompts help you draft objectives?

1. Objectives from strategy

Start from what the business needs this period, not from the team's task list.

Here is our strategy for {{period:Q1 2027}} and last quarter's results:

Strategy:
{{strategy}}

Results:
{{results}}

Suggest 5 candidate objectives for {{team:the growth team}}. Each should
be qualitative, memorable, and clearly linked to a line in the strategy
(quote it). Then recommend the 2 or 3 that matter most and say why the
others can wait.

2. Objective wording check

Short, clear objectives get remembered and repeated.

Rewrite each objective below so it is short, inspiring and specific
about what changes. Keep the meaning. No numbers in the objective (those
go in key results), no jargon, and no verbs like "support" or "help"
that hide the outcome. Give 2 versions of each.

{{objectives}}

Which prompts write key results?

3. Key results that measure outcomes

The most common OKR mistake is a key result that is really a task.

Write 3 to 5 key results for this objective: {{objective}}

Metrics we can measure today, with current values:
{{metrics}}

Each key result must be a measurable outcome with a baseline and a
target, in the form "from X to Y by {{deadline:31 March}}". No tasks or
launches as key results; put those under "initiatives" instead. If a
metric I'd need isn't in my list, say so rather than inventing a number.

4. Turn tasks into key results

For a draft full of "launch", "build" and "run".

These draft key results describe activities, not outcomes:

{{draft_krs}}

For each, ask "what result do we want from doing this?" and rewrite it
as a measurable outcome. Keep the original activity as an initiative
under the new key result. Flag any where the outcome can't be measured
yet and suggest how to start measuring it.

5. Find the baselines you need

Run this before committing. A target without a baseline is a guess.

Here are our draft OKRs:

{{okrs}}

For each key result, tell me: the baseline value we need, where it would
come from ({{tools:our analytics tool, CRM and support desk}}), how often
we can measure it, and whether we can measure it at all today. List the
key results we should rewrite because the data doesn't exist.

Which prompts check OKRs before you commit?

6. Critique a draft set

A skeptical review before the planning meeting.

Review these OKRs for {{team:the growth team}} as a tough but fair
coach.

{{okrs}}

Check each one for: objective that is really a metric, key result that
is really a task, missing baseline, a target that is either trivial or
impossible, key results that could be hit while the objective fails, and
too many OKRs overall. Quote the problem and suggest a fix.

7. Label committed versus aspirational

What Matters says "a committed OKR is expected to have a passing grade," while aspirational ones are "sometimes called stretch goals." Mixing them without labels causes confusion at scoring time.

Label each OKR below as committed or aspirational, and explain why in
one line. Committed means we expect to fully deliver it. Aspirational
means partial progress is still a good result.

For committed ones, check the target is realistic given:
{{capacity}}

For aspirational ones, check the target is truly ambitious.

OKRs:
{{okrs}}

8. Cut to what matters

When every team lead adds "just one more" objective.

We have {{count:7}} objectives and {{kr_count:26}} key results for
{{period:Q1 2027}}. That's too many to focus on.

{{okrs}}

Rank them by impact on {{goal:revenue growth}} and by effort. Recommend
a set of no more than 3 objectives with 3 to 5 key results each. Say
what we drop and what happens if we drop it.

Which prompts align OKRs across teams?

9. Team OKRs from company OKRs

Alignment without copying the company OKRs word for word.

Company OKRs for {{period:Q1 2027}}:
{{company_okrs}}

My team: {{team:the growth team}}, which owns {{scope:signup, onboarding and activation}}.

Draft 2 or 3 team OKRs that contribute to the company OKRs. For each,
say which company key result it moves and how. Only use metrics my team
can influence directly. Flag company OKRs my team can't affect.

10. Spot conflicts between teams

Two teams pulling in opposite directions is easy to miss on separate pages.

Here are OKRs from several teams:

{{all_team_okrs}}

Find: key results that conflict (one team's win hurts another's),
duplicates where two teams own the same number, dependencies where one
team needs another's help, and company goals no team covers. Return a
table with the issue, teams involved and a suggested fix.

Which prompts help run and score OKRs?

11. Weekly check-in

Keeps OKRs alive between planning and grading.

Here are our OKRs and this week's numbers and notes:

OKRs:
{{okrs}}

This week:
{{updates}}

For each key result: progress versus where we should be by now, a
confidence level (on track, at risk, off track) with the reason, and the
one action that would help most next week. Keep it under {{words:250}} words.
Flag anything that has been at risk for more than {{weeks:2}} weeks.

12. Initiatives to move a stuck key result

When a number hasn't moved for weeks.

This key result is stuck: {{key_result}}
Current value: {{current}}, target: {{target}}, weeks left: {{weeks_left:5}}

What we've tried: {{tried}}

Suggest 5 initiatives that could move it in the time left, ranked by
expected impact and effort. For each, say what early signal would show
it is working within 2 weeks. Tell me honestly if the target is now out
of reach.

13. End-of-quarter scoring

Scores each key result the same way every time. This prompt uses the 0.0 to 1.0 scale that What Matters describes for Google's method.

Score each key result below from 0.0 to 1.0 based on the final value
versus the baseline and target. Show the calculation. Note which OKRs
were committed and which aspirational, and comment on the score in that
light. Then give each objective an average score.

{{okrs_with_final_values}}

14. Retrospective and next quarter

Turns the score into better OKRs next time.

Here are our scored OKRs for {{period:Q1 2027}} and team notes:

{{scored_okrs}}

{{notes}}

Tell me: which targets were set too low or too high and why, which key
results measured the wrong thing, and what we learned about our
capacity. Then draft OKRs for next quarter that carry forward what still
matters and fix those problems. Quote the evidence from the notes.

What does a good OKR look like?

A good objective says what changes, in words people remember. A good key result is a number with a baseline, a target and a date. Here is the difference in practice:

The weak column mixes a metric into the objective and lists tasks as key results. The better column keeps the tasks (new onboarding, experiments, docs) as initiatives that serve the key results.

How do you keep AI from writing generic OKRs?

Give it your numbers. A model that sees your real activation rate, churn or pipeline will write key results you can track. A model that sees only "growth team" will write ones you can't.

Tell it to say when data is missing. Anthropic's guide to reducing hallucinations recommends giving the model explicit permission to admit uncertainty, which stops it from inventing a baseline.

Keep the decision human. The model drafts and critiques; the team agrees what to commit to, because they are the ones who will be measured.

Make your favorite one a skill

If you run the same critique or check-in prompt every quarter, turn it into a skill with your OKR rules built in, so Claude, Claude Code or Codex applies them every time. Paste it into the free SKILL.md generator. For planning the work behind the key results, see the project management prompts.

To keep these ready for planning week, save them in Kipt with their {{blanks}}, tag them by stage (draft, check, score), and press ⌘K to find one. Your library lives as Markdown in a private GitHub repo you own.