| name | review-okrs |
| description | Use when reviewing, critiquing, or improving OKRs for a product or delivery team. Triggers on requests to review OKRs, check if key results are measurable, improve KR quality, or validate an OKR set before a planning cycle. |
review-okrs
Overview
The user is part of the ACIL team at carsales. Objectives are fixed — do not create, rename, or reword them. The user will provide KRs (sometimes delivery-focused). Your job is to:
- Map each KR to the correct ACIL objective
- Apply the pass/fail quality test
- Rewrite every failing KR with a concrete outcome metric
Never invent objectives. Never hallucinate KR context. If a KR cannot be confidently mapped, say so and ask.
ACIL Objectives (Fixed — Do Not Modify)
| # | Objective |
|---|
| O1 | To increase participation and engagement across AG Tools and ecosystems to support JTBD |
| O2 | To improve dealer productivity and platform efficiency |
| O3 | To expand revenue and platform growth |
| O4 | To expand carsales ecosystem through new partnerships |
Mapping logic:
- KRs about dealer usage, adoption, feature engagement → O1
- KRs about dealer workflows, time saved, reduced friction, data quality → O2
- KRs about new revenue, GMV, monetisation, paid features, upsell → O3
- KRs about integrations, third-party data, new partners, ecosystem APIs → O4
- If a KR could fit multiple objectives, choose the primary one and note the secondary
The 4-Question Test (run on every KR)
- Does it have a number? No number = not a KR.
- Is it an outcome or an output? Outputs describe what you ship. Outcomes describe what changes as a result.
- Does it have a baseline and a target? "Increase X" is not measurable. "Increase X from 40% to 65%" is.
- Can you complete it without achieving the objective? If yes, it's an output.
Red Flag Words → Automatic Fail
If a KR contains any of these words, it is an output disguised as a result:
launch · ship · deliver · deploy · implement · build · complete · release · migrate (as the end goal) · ensure · improve (without a number)
Exception: "migrate" is acceptable as a KR when paired with a quality metric (e.g. "Migrate 100% of records with < 1% error rate").
Objective Quality Check
A strong objective is:
- Qualitative and aspirational — describes the change in the world, not the work to be done
- User or business anchored — answers "for whom?" or "so that what?"
- Not a project name — "Complete prospect migration" is a project. "Dealers have accurate, unified customer records to act on" is an objective.
Output Format
Group KRs under their mapped ACIL objective. For each:
### O[1–4]: [ACIL Objective text]
**KR: [original KR text]**
Mapped to: O[N] — [one-line reason]
Quality: PASS / FAIL
Issue: [specific problem, or "none"]
Rewrite: [concrete replacement with baseline → target, or "no change needed"]
After all KRs, add:
## Summary
- [N] KRs pass · [N] fail
- Mapping: O1 [n] · O2 [n] · O3 [n] · O4 [n]
- Patterns: [e.g. "All KRs are outputs — no outcomes measured"]
- Top fix: [the single most important change]
- Unmapped: [any KRs you couldn't confidently assign — flag for user to confirm]
Rewrite Rules
When rewriting a failed KR:
- Include a metric, a baseline (use
[TBC] if unknown), and a target
- Use the format:
[Metric] from [baseline] to [target] by end of quarter
- If the output-based KR maps to an obvious outcome, use that. If it doesn't, flag it: "No clear outcome for this output — confirm what success looks like."
Examples:
| Original (Fail) | Rewrite (Pass) |
|---|
| Ship call transcript feature | % of completed calls with enrichment summary saved to profile: [TBC] → 60% |
| Launch after-hours auto-response | Median response time to after-hours enquiries: [TBC] → reduce by 70% |
| Ensure data quality is good | Post-migration mandatory field completion rate: [TBC] → ≥ 95% |
| Increase dealer satisfaction | Dealer CSAT score: [TBC] → [target] (define measurement method first) |
Structural Rules
| Check | Pass criteria |
|---|
| Number of objectives | 2–3 per team per quarter |
| KRs per objective | 2–4 (max 5) |
| Initiatives listed as KRs | Flag and move to initiatives list |
| All KRs traceable to objective | Each KR should visibly move the objective needle |
| Baselines confirmed | Flag [TBC] items as blockers before quarter starts |
Common Mistakes
- Objective = project name. Fix: reframe around the outcome of completing the project.
- KR = "100% complete" on a task. Fix: measure fidelity or impact, not binary completion.
- Satisfaction KR with no measurement method. Fix: name the survey, score, or signal before writing the target.
- Overlapping KRs. If two KRs both describe the same feature from different angles, consolidate or clarify.
- Too many KRs. More than 4 per objective = loss of focus. Cut the weakest.
