OKR Design Facilitator

---
name: okr-design-facilitator
description: Use when designing a company or team OKR set for the quarter — before sprint planning or roadmap locking. Triggers on "OKR", "quarterly goals", "Objectives and Key Results", "what should we optimize for this quarter", or "help me write our Q[X] goals".
---

# OKR Design Facilitator

OKRs fail in one of two ways: the Objective is too vague to align on, or the Key Results measure activity instead of outcome. This skill produces a properly formed OKR set that passes the Doerr spec checks — and a confidence-scoring cadence to know whether you're on track before the quarter ends.

## Required inputs

1. **Company or team context** — stage, headcount, current top metrics
2. **The Objective scope** — company-level, team-level, or individual OKR?
3. **What you're trying to move** — the outcome or metric that matters most this quarter
4. **Current baseline** — what does that metric look like today?
5. **Last quarter's OKRs** (if any) — to check for carryover vs. intentional change

## What a properly formed OKR looks like

### The Objective

- **Qualitative and ambitious** — states direction, not a number
- **A team member can explain it without looking at a doc** — "reduce churn before it becomes structural" passes; "achieve a net retention improvement that supports our ARR growth trajectory" does not
- **Time-bounded to the quarter** — not an evergreen mission, not a 3-year vision

### Key Results (3–5 per Objective)

Each KR must pass three tests:

**1. Outcome, not output.**
"30-day retention rises from 61% to 72%" is an outcome — it measures what happened because of the work.
"Ship 3 onboarding experiments" is an output — it measures what the team did, not what changed.

**2. Independently measurable.**
You can check each KR's number without checking the others first. If KR2 depends on KR1 being measured, they're entangled.

**3. Defining.**
Together, the KRs should fully define what "we achieved the Objective" means. If you hit all KRs but the Objective still doesn't feel achieved, the KRs are wrong.

## Common failure patterns — catch before finalizing

**Output KRs disguised as outcomes**

Bad: "Launch weekly onboarding emails"
Good: "30-day email open rate on onboarding sequence reaches 38%"

Bad: "Conduct 20 customer interviews"
Good: "Customer-reported clarity of value prop on activation survey improves from 3.1 to 4.2/5"

**Vague Objectives**

Bad: "Improve customer experience"
Good: "Make our onboarding reliable enough that customers reach first-value moment in their first session"

**KRs that all move from the same lever**

If three of five KRs all change when you adjust the same variable, you have one KR written three ways. Collapse them.

**KRs without baselines**

A KR with no baseline ("NPS improves significantly") is unmeasurable. Every KR needs a current state and a specific target.

## Weekly confidence scoring

Each Friday, score each KR 0–10:
- 10 = already done
- 5 = on track, no changes needed
- 0 = definitely missing

Escalation triggers:
- Score < 4 by week 4: bring to team for plan change or scope adjustment, not just more effort
- Two consecutive weeks of flat score: KR is blocked, needs active intervention
- Score > 8 by week 8: consider stretching the target; it was too easy

## Output structure

\`\`\`
## Q[X] OKR — [Company / Team Name]

### Objective
[One sentence, qualitative, ambitious, quarter-scoped]

### Key Results

| # | Key Result | Baseline | Q[X] Target | Measurement method |
|---|---|---|---|---|
| 1 | [Outcome statement] | [Current metric] | [Target metric] | [How / when measured] |
| 2 | | | | |
| 3 | | | | |

### Spec check
- [ ] Each KR is an outcome, not an output or activity
- [ ] Each KR is independently measurable
- [ ] Together, the KRs define "done" for the Objective
- [ ] Every KR has a concrete baseline and a specific target

### Weekly confidence scoring template
KR1: [score]/10 — [one-sentence status]
KR2: [score]/10 — [one-sentence status]
Escalation needed? Y/N — [reason if Y]
\`\`\`

## Tone

- Push back on output KRs. "Conduct N interviews" is an activity — find the outcome it's supposed to drive.
- Push back on baselines that don't exist. "We don't track that yet" is itself a KR candidate: "Establish baseline measurement of X by end of week 3."
- If the team has two Objectives they can't decide between, don't blend them into one vague Objective. Run two trees and pick the one with the shorter feedback loop.