Use this when a PM needs to define or improve objectives and key results (OKRs) for a quarter, release, or initiative, and wants KRs that are outcome-based, measurable, realistic, and decision-useful.
Process
Step 0: Confirm readiness before drafting OKRs
Check for common failure modes before writing objectives:
- Is strategy clear enough to explain why this outcome matters now?
- Is there at least directional instrumentation for baseline metrics?
- Is the team trying to use OKRs for everything (instead of priority outcomes)?
- Is there a cross-functional group ready to own and review progress?
If readiness is weak, output:
Readiness gapsFast fixes this cycleWhat to defer to next cycle
Use reference/okr-readiness-and-anti-patterns.md.
Step 1: Gather context before writing goals
Ask for:
- Time horizon - What period are these OKRs for? (e.g., Q2, next 12 weeks)
- Outcome area - What user or business problem must improve?
- Starting point - What baseline metrics exist today?
- Constraints - Team size, dependencies, fixed deadlines, or platform limits
- Strategic intent - Why does this matter now?
- Known non-goals - What is explicitly out of scope?
- Alignment context - Which higher-level OKR(s) should this support?
- Dependency context - Which teams must contribute, and by when?
If baseline data is missing, ask for a fast proxy (last month, last release, sampled data) and mark confidence.
Step 2: Draft objective candidates (motivation first)
Generate 2-3 objective options, then help the user choose one. Use strong/weak calibration patterns from reference/okr-example-pack-good-vs-weak.md.
Each objective must:
- Start with an action verb (
Delight,Reduce,Enable,Increase,Simplify) - Describe an outcome, not a deliverable
- Connect user value and business value
- Be short enough to remember and repeat
- Be challenging but still plausible for the team
Use this check:
- Would this still make sense if specific features changed?
- Would the team feel energized reading this every week?
- Is this specific enough to guide decisions?
Step 3: Draft key results (measurement discipline)
Create 3-5 KRs for the chosen objective. For each KR, use:
(Metric verb) (metric name) from (baseline) to (target) by (date/timebox).
Example format:
Increase weekly activated teams from 38% to 52% by June 30.
KR mix rules:
- Keep the set to 2-5 KRs
- Include at least one user outcome KR
- Include at least one business or system impact KR
- Prefer a leading + lagging balance
- Avoid task completion as a KR (
launch,ship,build,write) - Include at least one counter-measure ("do not harm" guardrail)
- Label each KR as
CommittedorAspirational
If ideal metrics are expensive or unavailable, use a temporary proxy metric and state why it is a valid proxy. When examples are requested, reuse formats from reference/okr-example-pack-good-vs-weak.md.
Step 4: Stress-test realism and controllability
For each KR, validate:
- Measurable - clear metric, baseline, target, and timebox
- Realistic stretch - ambitious but plausible given constraints
- Controllable - team actions can materially influence the metric
- Evidence-backed - target rationale is explicit (historical trend, benchmark, experiment)
- Data availability - source and owner are known
Use the scorecard in reference/kr-quality-rubric.md. Flag low-scoring KRs and rewrite them. Then run the full checklist in reference/okr-health-checklist.md and identify red flags in:
Well definedWell alignedWell tracked
Step 5: Produce the final OKR package
Return this structure:
## Objective
(One selected objective)
## Key Results
| KR | Type (Committed/Aspirational) | Baseline | Target | Timebox | Metric source | Confidence |
|----|-------------------------------|----------|--------|---------|---------------|------------|
| KR1 | ... | ... | ... | ... | ... | High/Med/Low |
## Counter-measure (Do No Harm)
- (Guardrail metric and threshold)
## Why these KRs matter
- (How each KR ties to objective and strategy)
## Alignment map
- Parent OKR(s): (L1/L2 that this supports)
- Peer dependencies: (Teams, expected contribution, due date)
## Initiatives (not KRs)
- (Top bets the team will run to influence KRs)
## Risks and dependencies
- (Critical risks or external dependencies)
## Review cadence
- Weekly: (leading indicators)
- Monthly: (lagging indicators + decision checkpoint)
- Quarterly: (target reset, scope shifts, or KR retirement)
Always separate initiatives from KRs.
Step 6: Set operating cadence (track, learn, adjust)
Run monthly OKR progress reviews with PM, design, engineering, and key stakeholders.
Use the RCAR loop:
Reportlatest measure (fresh data with clear owner)Compareagainst prior period and trendAssessdirection vs. target (positive/flat/negative)Respondwith explicit next decision (ramp up, pivot, pause, or retire)
Use reference/okr-operating-cadence.md.
Step 7: Plan sequencing across quarters (optional, when requested)
When the user asks for annual or multi-quarter planning, build an outcome sequence across problem discovery, solution discovery, and delivery.
Use reference/okr-sequencing-and-levels.md.
Step 8: Facilitate a 60-minute OKR workshop (optional, when requested)
When the user asks to run a live workshop, follow the script in reference/okr-workshop-60-min-script.md. Use the minute-by-minute agenda and output template verbatim unless the user asks for a different duration.
Step 9: Review and tighten
Ask:
- Which KR feels hardest to influence, and why?
- Which KR might be a vanity metric?
- What decision would we make if KR progress stalls by week 4?
- Is anything important missing because it is hard to measure?
Revise and finalize.
Output location
Present the final objective and KRs as formatted text in the conversation so the PM can copy them into Jira, Notion, Productboard, or planning docs.
Example Output
Input
- Company & team: Meridian Health — Care Navigation product team (8-person squad: 2 engineers, 1 designer, 1 data analyst, 1 PM, 3 clinical ops liaisons)
- Time horizon: Q3 2025 (July 1 – September 30)
- Outcome area: Members who start a care navigation session are not completing it — 61% drop off before reaching a confirmed next step (e.g., specialist referral, appointment booked, care plan acknowledged)
- Baseline metrics: Session completion rate: 39%; median time-to-confirmed-next-step: 11 days; NPS from post-session survey: 28
- Strategic intent: Meridian's 2025 company OKR is to reduce avoidable ER utilization by 15%; internal data shows members who complete navigation sessions have 22% fewer ER visits within 90 days — so session completion is a key lever
- Constraints: No new third-party integrations until Q4 due to InfoSec review freeze; clinical ops liaisons are shared with two other squads
Output (abbreviated)
Objective
Make care navigation a journey members finish, not a process they abandon.
Why this over alternatives: Two other candidates were considered — one focused on enrollment volume (leading the wrong outcome) and one on clinician productivity (misses the member experience angle). This framing keeps the team centered on completion as an outcome, survives feature changes, and directly connects to the parent company OKR on ER reduction.
Key Results
| KR | Type | Baseline | Target | Timebox | Metric Source | Confidence |
|---|---|---|---|---|---|---|
| KR1: Increase session completion rate (member reaches confirmed next step) from 39% to 58% | Committed | 39% | 58% | Sep 30 | Care nav platform event log (owned: data analyst) | Med |
| KR2: Reduce median time-to-confirmed-next-step from 11 days to 6 days | Committed | 11 days | 6 days | Sep 30 | CRM milestone timestamps (owned: clinical ops) | Med |
| KR3: Increase post-session NPS from 28 to 42 | Aspirational | 28 | 42 | Sep 30 | In-app survey, n ≥ 200/month (owned: PM) | Low — survey response rate currently 18%; treat as directional proxy until sample improves |
| KR4: Achieve ≥ 70% of completing members taking a confirmed action within 48 hrs of session close | Aspirational | ~31% (sampled) | 70% | Sep 30 | Platform event log + CRM cross-match (owned: data analyst) | Low — baseline is sampled, not instrumented; instrument by July 15 |
Counter-measure (Do No Harm)
- Guardrail: Member-reported feeling-of-pressure score (post-session survey Q7) must not exceed 2.1/5.0. Completing a session by nudging members into unwanted care pathways undermines clinical trust and plan liability. If score exceeds threshold in any two consecutive weeks, pause personalization experiments and escalate to clinical ops lead.
Why These KRs Matter
- KR1 is the primary lagging outcome — directly predicts the ER reduction Meridian's L1 OKR requires. Each 5-point completion gain is estimated to prevent ~120 unnecessary ER visits per quarter based on historical cohort data.
- KR2 is a leading indicator. Shorter time-to-next-step correlates with completion; if members wait more than 7 days for a clear action, abandonment spikes to 74% (internal Q1 analysis).
- KR3 captures qualitative signal that the team is improving experience, not just engineering funnel pressure. An NPS below 30 suggests members feel lost or coerced — a clinical and brand risk.
- KR4 operationalizes "completion" beyond a checkbox — confirmed action within 48 hours distinguishes genuine navigation from passive acknowledgment.
Alignment Map
- Parent OKR (L1): Reduce avoidable ER utilization by 15% (Meridian 2025 company objective)
- Supporting L2: Improve care navigation effectiveness so that completed sessions drive downstream health actions (this objective)
- Peer dependencies:
| Team | Expected Contribution | Due Date |
|---|---|---|
| Data & Analytics | Instrument KR4 baseline in event log | July 15 |
| Clinical Ops (shared) | Audit and standardize "confirmed next step" taxonomy across 4 session types | July 22 |
| Member Experience (design) | Deliver friction audit of 3 highest drop-off moments in current flow | July 8 |
Initiatives (Not KRs)
These are the bets the team will run to move the KRs — not KRs themselves:
- Redesign the session handoff screen to surface one clear next action (reduces choice paralysis at the primary drop-off point, minute 7–9)
- Build a 48-hour SMS/push reminder for members who leave without a confirmed next step
- Introduce asynchronous care plan acknowledgment so members on mobile don't require a live callback
- Run a cohort analysis to identify whether drop-off clusters by condition type, session length, or time of day — inform Q4 personalization work
Risks and Dependencies
- InfoSec freeze: No new integrations until Q4 limits automated appointment booking. KR2 target may require manual outreach workarounds through clinical ops liaisons — capacity risk given shared team.
- KR4 instrumentation gap: If data analyst cannot instrument the 48-hour action metric by July 15, KR4 becomes untrackable for the first 6 weeks. Contingency: use sampled CRM spot-check biweekly until live.
- Survey response rate: NPS (KR3) is directional only until response rate clears 25%. Do not make resourcing decisions based on KR3 alone in July.
- Clinical ops bandwidth: Two other squads share the three liaisons. If taxonomy audit (dependency above) slips past July 22, KR1 and KR4 definitions become inconsistent across session types.
Review Cadence
- Weekly (Mondays, 20 min): KR2 (time-to-next-step trend), drop-off rate by session stage, guardrail score — owned by PM, reported in shared Notion dashboard
- Monthly (1st Thursday, 60 min): Full RCAR review across all four KRs; explicit decision: ramp up, pivot, or pause each initiative; include clinical ops lead and one engineering lead
- Quarterly (Sep 24, pre-QBR): Final KR attainment; decide which KRs graduate to BAU metrics vs. carry into Q4; document what the team learned about the drop-off mechanism for Q4 sequencing
Step 9 — Stress-Test Prompts for PM to Resolve
- KR4 is the hardest to influence directly — the team controls the reminder and handoff design, but member follow-through depends on clinical availability. What's your contingency if confirmed actions stall despite completed sessions?
- KR3 (NPS) risks becoming a vanity metric if response rate stays below 20%. Would you accept replacing it with a single-question effort score (CES) that is easier to collect passively?
- If KR1 (completion rate) is flat by week 4, the default decision should be: pause the handoff redesign experiment and run qualitative sessions with 10 members who dropped off in the prior 2 weeks — agree?
- Is there anything important missing because it's hard to measure — e.g., whether members who complete navigation actually follow through with the specialist referral vs. just acknowledging it?