Prediction Coverage Frontier
Understand which revenue movements learning can predict and which data gaps still limit coverage.
Prediction Coverage Frontier is a support-oriented learning report that explains where Prevenue has mature outcome evidence, where learned signals are operationally strong enough, and where the model is still blind. It is generated from account feature snapshots, mature outcome windows, learned pattern candidates, readiness metadata, mapping coverage, account matching, and Event KPI evidence. It does not expose raw event payloads.
Use it after a learning phase completes to answer which outcome groups are covered:
| Outcome group | Examples |
|---|---|
| Expansion | Expansion and top-up outcomes. |
| Contraction | Downgrades and lower recurring revenue. |
| Churn | Lost paid subscriptions. |
| Conversion | Trial or free-to-paid conversion. |
| Payment recovery | Failed-payment recovery outcomes. |
How coverage is counted
Decision-ready coverage only counts mature outcomes matched by learned signals that passed validation, confidence, freshness, and fallback gates. Directional and exploratory correlations are shown separately so they can inform investigation without inflating coverage claims.
Each outcome group can include:
- Observed mature outcomes.
- Outcomes covered by operationally eligible learned signals.
- Directional matches that are report-only.
- Exploratory matches that need more evidence.
- Uncovered outcomes.
- Evidence quality.
- Blind spots and instrumentation recommendations.
Blind spots
Blind spots explain why a movement is not covered yet. Common reasons include low mapping coverage, weak account matching, missing feature families, sparse outcome samples, thin or unavailable Event KPI evidence, and weak statistical evidence.
Use blind spots to decide what to instrument next. The most useful improvements are usually confirming event mappings, sending stable account identifiers, adding value metrics, and classifying value, friction, sentiment, and growth-intent events.
Interpreting the report
Treat decision-ready coverage as the only coverage eligible for bounded Funnel decisions. Treat directional and exploratory sections as investigation material until learning has enough mature outcomes and validation evidence.