| Term | Meaning |
|---|---|
| Project | The brand and measurement configuration you are reading. |
| Run | One measurement execution with its own ID and completion state. |
| Query or prompt | A question submitted for measurement. One query can produce responses from more than one platform. |
| Response | A whole answer from an AI platform. Coverage uses response-level presence. |
| Statement | A returned section of answer text, such as a paragraph, list item, or table row; it can contain several sentences. |
| Occurrence | One recorded mention match. Several can occur in the same statement or response. |
| Entity | A measured brand term or other recognized name. |
| Bundle | A configured grouping of entity terms under a brand label. Do not assume every metric can be added across bundle members. |
| Domain | A website host used for source-level reporting. |
| URL | A specific source page. |
| Subject | The brands or sources selected for return, separate from the population filters. |
| Filter | A restriction on the measurement scope. |
| Group | A dimension by which returned results are split. |
| Aggregation level | Whether results describe domains, URLs, entities, or bundles, depending on the tool. |
| Confidence interval | The returned range expressing uncertainty around an estimate or difference. |
| Signal-to-noise ratio | The reported indicator of signal separation relative to uncertainty. Interpret it for the returned domain or URL scope. |
| Convergence | Both the ranking and signal sufficiency checks have passed for the reported scope. |
| Decision-grade | A result with suitable scope, stability, and uncertainty information for the decision being considered. It is not a guarantee of an outcome. |
| Significant comparison | A returned contrast whose test establishes a direction at its reported confidence level; retain any correction limitations. |
| Segment | A returned evidence tier: elite, core, established, or visible. Preserve the server label; do not reconstruct the classification. |
visible segment | The weakest evidence tier. Do not present it alone as a confirmed competitive finding or proof of absolute absence. |
stable estimate | A row-level quality label; different from run-level convergence. |
| Groundedness | Citation support for the measured brand mention statements. It is not a universal fact-checking guarantee. |
| Snapshot | A stored period of Google data. Its resolved window may differ from the requested date range. |
| MCP | Model Context Protocol: the connection through which compatible clients discover and call IQRush tools. |
Reference
Developer glossary
Use these terms consistently in your application and reports.
⌘I