Product
Metrics & success criteria
How we define done for product work — activation, retention, trust, impact
Metrics & success criteria
Metrics exist to decide, not to decorate decks. Prefer a few honest numbers over vanity charts.
North-star style outcomes
| Outcome | Signals (examples) |
|---|---|
| Receipts are usable | Time-to-find receipt; successful return/warranty lookup |
| Shoppers trust the vault | Auth completion; bind success; low support “missing receipt” volume |
| Daily habit forms | WAU / sticky sessions into Bills or Analysis |
| Offers are real | View → redeem rates (not just impressions) |
| Impact is credible | Paper saved / CO₂ fields only when sourced |
Exact instrumentation lives in product analytics configs per surface — document definitions here when they stabilize; do not invent fake targets in MDX.
Definition of done (feature)
A feature is not done until:
- The shopper job works on the happy path and a named failure path.
- The owning surface is correct (no half-portals).
- Privacy / auth boundaries are respected.
- A metric or qualitative kill criterion is named.
- Design follows Design language where UI ships.
Guardrail metrics
| Risk | Watch for |
|---|---|
| Auth friction | Drop-off between open → authenticated RootTabView |
| Bind failure | Incomplete UCI / last4 attempts; retry loops |
| Data distrust | Spike in “receipt wrong / missing line” reports |
| Greenwashing | Eco UI without backend-backed fields |
How PMs should write goals
Prefer:
Increase successful receipt detail opens from Home within 7 days of first bind.
Avoid:
Make Analysis more engaging.