define-okrs

name	define-okrs
description	Define OKRs (Objectives and Key Results) for a product initiative, quarter, or team. Enforces baselines, measurement methods, and the 70% completion target philosophy.
argument-hint	[initiative, quarter, or team to define OKRs for]
user-invocable	true
allowed-tools	Read, Write, Edit, Glob, Grep

Define OKRs for $ARGUMENTS.

Follow every step below. OKRs without baselines and measurement methods are useless — this process ensures every OKR is actionable.

---

Step 1: Understand the Context

Before writing any OKRs, establish:

1. Time horizon: What period do these OKRs cover? (Quarter is standard. Shorter for sprints, longer for annual planning.)
2. Scope: Is this for a product, a team, a specific initiative, or the whole company?
3. Parent OKRs: Do these ladder up to a higher-level objective? If so, read and reference it. Every team OKR should visibly connect to a company or department objective.
4. Current state: What happened last period? Which OKRs were hit, missed, or abandoned? This context prevents repeating failed approaches.

---

Step 2: Write Objectives

Each objective must meet ALL of these criteria:

Objective Rules

1. Qualitative, not quantitative. Objectives describe the desired future state in human language. Numbers go in Key Results, never in Objectives.

   • Bad: "Increase NPS from 32 to 50"
   • Good: "Become the product customers actively recommend to peers"

2. Inspiring and memorable. If you cannot remember the objective without looking it up, it is too bland. Objectives should create energy.

   • Bad: "Improve platform reliability"
   • Good: "Make downtime a distant memory for our customers"

3. Time-bound by the OKR period. The objective should be achievable within the quarter (or whatever period is set). Multi-quarter objectives belong one level up.

4. Outcome-oriented, not activity-oriented.

   • Bad: "Launch the new onboarding flow" — that is an activity
   • Good: "Make new users productive on day one" — that is an outcome

5. Limited to 2-4 objectives per team per quarter. More than 4 means nothing is truly a priority. If you have more than 4 candidates, force-rank and cut.

Anti-Patterns for Objectives

• Business-as-usual dressed up as an objective ("Keep the site running")
• Objectives that are actually key results ("Reduce churn by 20%")
• Objectives so broad they could mean anything ("Be world-class")
• Objectives that no one would disagree with ("Deliver value to customers")

---

Step 3: Write Key Results

Each objective gets 3-5 Key Results. Fewer than 3 means the objective is not well-defined. More than 5 means the objective is too broad — split it.

Key Result Rules

1. Quantitative with a specific number. Every KR has a metric, a baseline, and a target.

   KR: Reduce median time-to-first-value from 14 minutes to 4 minutes
        ↑ metric                              ↑ baseline    ↑ target
   

2. Baselines are mandatory. A target without a baseline is a guess, not a goal.

   • If the baseline is known, state it with the data source: "Current: 14 min (measured via Mixpanel, Q4 average)"
   • If the baseline is unknown, the FIRST Key Result must be: "Establish baseline measurement for [metric] by [date]" — this is the only acceptable non-numeric KR

3. Outcomes, not outputs.

   • Output (bad): "Ship 3 new features"
   • Outcome (good): "Increase weekly active usage from 45% to 65%"
   • The distinction: outputs are things you control. Outcomes are the results of those things. OKRs measure outcomes.

4. 70% completion = success. Targets should be ambitious enough that achieving 70% represents a strong result. 100% means you set the bar too low. Below 50% means the target was unrealistic or the approach was wrong.

5. No binary KRs. "Launch feature X" is binary — it is either done or not. Convert to: "Feature X adopted by 30% of eligible users within 4 weeks of launch."

6. Include the measurement method. For each KR, state exactly how it will be measured:

   • What tool or data source?
   • What query or report?
   • Who is responsible for measuring?
   • How often is it measured?

Leading vs. Lagging Indicators

Every set of Key Results should include both:

Leading indicators — things that change quickly and predict future outcomes:

• Feature adoption rate (first week)
• Funnel conversion at a specific step
• Support ticket volume for a specific category
• User engagement frequency

Lagging indicators — things that confirm the outcome but take time to materialise:

• Retention rate (needs 30-90 days)
• Revenue impact (needs a full billing cycle)
• NPS or satisfaction score (needs a survey cycle)
• Churn rate (needs months to stabilise)

A good OKR set has at least one leading KR (so you can course-correct early) and at least one lagging KR (so you know if the outcome was real).

---

Step 4: Validate the OKR Set

Run every OKR through this checklist:

Objective Validation

• Qualitative — no numbers in the objective
• Inspiring — would motivate the team if read aloud
• Achievable within the time period — not multi-quarter
• Outcome-oriented — describes the future state, not activities
• Distinct — does not overlap significantly with another objective

Key Result Validation

• Has a specific metric with a number
• Baseline is stated (or first KR is to establish it)
• Target follows the 70% rule — ambitious but not absurd
• Measures an outcome, not an output
• Measurement method is documented (tool, query, frequency, owner)
• Not binary — has a spectrum of achievement
• Mix of leading and lagging indicators across the KR set

Set-Level Validation

• 2-4 objectives total (no more)
• 3-5 key results per objective
• KRs across objectives do not contradict each other
• The set covers both growth AND health (do not optimise one at the expense of the other)
• At least one KR is a guardrail metric (something that should NOT get worse while you pursue the others)

---

Step 5: Define the Scoring Framework

At the end of the period, each KR is scored on a 0-1 scale:

Score	Meaning
0.0	No progress
0.3	Some progress but fell well short
0.5	Made meaningful progress, hit roughly half the target
0.7	Strong result — this is the expected "good" outcome
1.0	Fully achieved — target may have been too easy

Team OKR score = average of all KR scores. A team score of 0.6-0.7 indicates healthy ambition and strong execution.

---

Output Format

# OKRs: [Team/Initiative] — [Period]

**Context:** [1-2 sentences on what happened last period and what is driving this period's focus]

**Parent objective:** [Reference to the company/department OKR this ladders up to, if applicable]

---

## Objective 1: [Qualitative, inspiring statement]

| KR | Metric | Baseline | Target | Measurement | Type |
|----|--------|----------|--------|-------------|------|
| KR1 | [Metric name] | [Current value + source] | [Target value] | [Tool, query, frequency] | Leading |
| KR2 | [Metric name] | [Current value + source] | [Target value] | [Tool, query, frequency] | Lagging |
| KR3 | [Metric name] | [Current value + source] | [Target value] | [Tool, query, frequency] | Guardrail |

**Why this objective:** [2-3 sentences on why this matters now]

**Key initiatives (inputs, not measured):** [What the team plans to do to move these KRs — listed for context, not as commitments]

---

## Objective 2: [Qualitative, inspiring statement]

[Same format]

---

## Risks and Dependencies
- [Risk or dependency that could affect achievement]
- [Risk or dependency that could affect achievement]

Write the output to a file: docs/okrs-[team-or-initiative]-[period].md.

Related Skills

• /coordinator:decompose-initiative — decompose the initiative into workstreams before defining OKRs. Decompose first (who does what), then define OKRs (how we measure success).