| name | kpi-architect |
| description | Use when the user wants to define KPIs, build a metrics framework, identify leading indicators, create a risk register, or set up governance triggers. Also use when the user mentions 'KPIs', 'metrics framework', 'leading indicators', 'lagging indicators', 'risk register', 'OKRs', 'metric governance', or 'war-gaming'. |
KPI Architecture
Expert knowledge for designing a metrics framework that drives decisions rather than generates reports.
The 2-3-2 Rule
The maximum practical KPI set for any team or product:
2 Leading Indicators — predict future performance (you control these)
3 Operational Metrics — measure current health (you monitor these)
2 Lagging Indicators — confirm past performance (you report these)
─────────────────────
7 metrics total
Why max 7: More than 7 metrics on a dashboard means no metric is a priority. When everything is a KPI, nothing is.
Leading Indicator Validation
A leading indicator is only valuable if it predicts the lagging outcome you care about. Validate before enshrining.
Test:
- Does a change in this metric precede a change in the lagging indicator?
- Is the lag time short enough to be actionable?
- Can the team actually influence this metric with their decisions?
If any answer is "no" or "uncertain" — it's a hypothesis, not a KPI. Label it as such and validate with 90 days of data.
Common leading indicator mistakes:
- Vanity metrics (page views, followers) that don't predict revenue or retention
- Metrics you can't influence (industry growth rate, competitor prices)
- Metrics with lag time > 90 days (by the time they lead, it's too late to act)
Metric Structure
For each KPI, define all five fields:
| Field | Description |
|---|
| Name | Plain English name |
| Formula | Exact calculation |
| Target | Threshold for good/at risk/critical |
| Frequency | How often measured |
| Owner | Who is responsible for moving it |
Example:
Name: Weekly Active Teams
Formula: Unique team_ids with ≥1 session in rolling 7 days
Target: Good: > 500 | At Risk: 400–500 | Critical: < 400
Frequency: Daily (reported weekly)
Owner: Head of Product
War-Gaming
War-gaming tests whether your strategy is robust to competitive attack.
3-step process:
- Name the attacker. Pick your most dangerous competitor or a plausible new entrant.
- Describe their move. What is the most damaging thing they could do in the next 90 days? (Price cut, feature launch, acquisition, partnership)
- Name your defense. What is your pre-planned response? If you have no response, that is a strategic gap.
Run war-gaming quarterly. Markets change. A defense that was solid 6 months ago may be obsolete.
Output format:
Attacker: [company or archetype]
Move: [specific action they could take]
Impact: [what metric this affects and by how much]
Defense: [your response and timeline to deploy it]
Gap: [yes/no — do you currently have this defense ready?]
Risk Register
A risk register captures the top threats to your plan. Keep it focused: top 3 risks only.
Format:
| Risk | Probability | Impact | Mitigation |
|---|
| Key engineer leaves | Medium | High | Cross-train second engineer on critical system |
| Competitor launches free tier | High | High | Accelerate feature differentiation roadmap |
| Regulatory change (GDPR enforcement) | Low | Medium | Compliance surface audit scheduled Q3 |
Scoring:
- Probability: High (> 50%), Medium (20–50%), Low (< 20%)
- Impact: High (threatens viability), Medium (affects quarterly target), Low (manageable)
Review cadence: Monthly. Remove risks that materialized (handle them in operations). Add new risks as they emerge.
Governance Triggers
Governance triggers define when a metric change is significant enough to require a decision or escalation — not a scheduled meeting.
Format:
Metric: [KPI name]
Threshold: [specific value or % change]
Trigger: [automated alert or meeting called]
Owner: [who reviews and acts]
Examples:
Metric: Weekly Active Teams
Threshold: < 400 (critical) for 2 consecutive weeks
Trigger: Emergency product review within 48 hours
Owner: Head of Product + CEO
Metric: Churn Rate
Threshold: > 5% monthly for any segment
Trigger: CS + Product sync within 1 week
Owner: Head of CS
Rule: Governance triggers fire on metric+threshold, not on calendar date. Weekly syncs that review the same green dashboard waste time.
OKR Integration
KPIs measure the health of the business. OKRs drive change. They work together:
- KPIs: Are we healthy? (Leading, operational, lagging — always on)
- OKRs: Are we improving? (Quarterly, time-boxed, 70% achievement = success)
Don't put your North Star Metric as an OKR Key Result — the NSM is a KPI. Use OKRs for the specific initiatives that move the NSM.
Common Rationalizations
| Rationalization | Reality |
|---|
| "We need more than 7 metrics to cover everything" | Coverage is not the goal. Decision-making is. More metrics = more noise = worse decisions. |
| "We'll add metrics as questions come up" | Metric sprawl is how dashboards die. Add only what has a defined owner and governance trigger. |
| "Our leading indicators are obvious" | Validate them against lagging outcomes with 90 days of data before treating them as facts. |
| "We review KPIs in our weekly meeting" | Scheduled reviews miss rapid changes. Governance triggers catch problems in 48 hours, not 7 days. |
Verification