Product Health Dashboard Spec

# Product Health Dashboard — Specification

**Owner:** Product Operations (definitions), Data Engineering (implementation)
**Refresh cadence:** Daily (data), weekly (digest)
**Audience:** Exec team + product team

---

## Five metric families

Every dashboard tile belongs to exactly one family.

1. **Adoption** — are customers turning the feature on?
2. **Activation** — are they reaching first value?
3. **Retention** — are they coming back?
4. **Satisfaction** — do they like it?
5. **Support load** — what is the operational cost of serving them?

---

## Tile spec — required fields per metric

| Field | Definition | Example |
| --- | --- | --- |
| Metric name | Plain English | Weekly active accounts |
| Family | One of the five | Adoption |
| Definition (English) | One sentence anyone can read | An account is "active" if any user logged in and performed a key action in the last 7 days |
| Source | Where the data comes from | Snowflake, table `events.user_actions` |
| SQL or query | Pseudocode or actual query | See SQL block below |
| Unit | Number, %, $, days | Count of accounts |
| Owner | Single person | Sam (PO) |
| Target | Numeric goal | 1,200 by end of Q |
| Last value | Most recent reading | 1,083 |
| Trend (4 weeks) | Sparkline or % change | +6.4% |
| Alert threshold | What triggers a slack notification | -10% week-over-week |

---

## Sample tile — Weekly active accounts (Adoption)

```sql
-- Definition: count of unique accounts with at least one
-- "key action" event in the trailing 7 days, ending Sunday.

SELECT
  date_trunc('week', event_at) AS week,
  count(distinct account_id)    AS weekly_active_accounts
FROM events.user_actions
WHERE event_name IN ('action.create', 'action.publish', 'action.export')
  AND event_at >= current_date - interval '12 weeks'
GROUP BY 1
ORDER BY 1;
```

- **Owner:** Sam (PO)
- **Target:** 1,200 by end of Q
- **Alert:** Page on-call PM if value drops 10% week-over-week
- **Last reviewed:** YYYY-MM-DD

---

## Standard tiles per family (start here, customize)

### Adoption
- Weekly active accounts
- Weekly active users
- % of new accounts that adopted feature X within 30 days

### Activation
- Time to first value (median, P75)
- % of signups that complete the activation milestone within 7 days
- Activation funnel conversion (signup → step 1 → step 2 → activated)

### Retention
- Week 4 retention (cohort)
- Net revenue retention (rolling 12-month)
- Logo churn (last 30 days)

### Satisfaction
- NPS (rolling 90-day)
- CSAT after support tickets
- Product-specific PMF score (Sean Ellis: % "very disappointed" if it went away)

### Support load
- Tickets per active account per month
- Median time to resolution
- Top 5 ticket categories by volume

---

## Governance

- Definitions changes require PO approval and a changelog entry.
- Owners review their tiles monthly. Stale tiles (no review in 60 days) get auto-flagged.
- Quarterly: drop any tile no one looked at in the last 90 days.

## Definition change log

| Date | Metric | Change | Why | Approved by |
| --- | --- | --- | --- | --- |
| | | | | |