The fastest way to make a metric useless is to let everyone own it.
"Customer satisfaction is down" becomes a company concern. Everyone nods. Support blames product quality. Product blames onboarding. Sales blames customer fit. Data says the survey methodology changed. Nobody is necessarily wrong. The metric has no operating owner, so the conversation becomes a tour of plausible explanations.
A real measurement system names the people who own definition, data, decision, and action. Those are not always the same person. Pretending they are is how reporting turns into theater.
A Metric Ownership Table
| Metric | Definition owner | Data owner | Decision owner | Action owner | Source of truth | Review cadence |
|---|---|---|---|---|---|---|
| Onboarding completion | Product Ops | Analytics | Head of Product | Activation PM | Warehouse model: onboarding_completion | Weekly |
| Sales velocity | RevOps | Data Eng | VP Sales | Segment sales leads | CRM + warehouse | Weekly |
| Net revenue retention | Finance | Analytics | COO | Customer Success | Finance close model | Monthly |
| Incident rate | Engineering | SRE | CTO | Service owners | Incident system | Weekly |
This table is boring. Good. Boring infrastructure prevents expensive confusion.
When Ownership Goes Wrong
Definition drift — teams use the same metric name for different things. "Churn" includes downgrades in one report and excludes them in another. Fix: maintain a metric dictionary and link definitions from the dashboard.
Data orphaning — the dashboard still exists, but the pipeline, owner, or source system changed. Fix: show freshness and owner metadata; archive dashboards that no longer have an owner.
Decision avoidance — everyone can explain the metric, but no one can say what happens when it moves. Fix: define thresholds and decision rights before the review.
Action diffusion — the decision is made, but execution lands in a group chat. Fix: one action owner, one due date, one follow-up venue.
