Evidence classes
Modeled gaps, observed analysis gaps and explicit requests are separate measures.
| Class | Meaning |
|---|---|
| Modeled gap | An uncovered need in a curated saved case. Its count measures case coverage. |
| Observed analysis gap | A completed real analysis that found no clean market fit. Raw owner language stays private. |
| Explicit request | A visitor actively asked for a missing market. |
Customer evidence reaches the partner surface only in scheduled groups of five. A missing customer cell can hold zero to four private records, so absence does not prove zero demand.
Modeled gaps
Modeled gaps come from curated, precomputed cases that completed the same public analysis stages. Across the maintained example set, they reveal recurring analytical shapes without representing a user, customer request or live analysis count.
Observed analysis gaps
A completed real analysis produces an observed gap when it has an owner-confirmed exposure and no clean funded relationship. Capture remains bounded to normalized dimensions such as risk type, time horizon and amount band, and one analysis contributes at most one observation per uncovered risk type.
Suggestions, clean fits, refusals and failed runs do not become observed gaps. Appearance during Detect alone never turns an analyst-inferred exposure into customer evidence.
Explicit requests
An explicit request records a deliberate action to ask for a missing market. Capture is idempotent so retries do not inflate the private evidence. Public display still waits for the same cohort rule as observed gaps.
Keep the units visible
| Evidence | Count unit | Safe interpretation |
|---|---|---|
| Modeled gap | Curated case | This need occurs in the maintained case set. |
| Observed gap | Released customer observation | At least this many completed analyses exposed the gap. |
| Explicit request | Released customer request | At least this many visitors deliberately requested the market. |