> ## Documentation Index
> Fetch the complete documentation index at: https://docs.iqrush.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Work through a visibility report

> This example shows how the same data becomes a clear report without losing its measurement context. It uses the Colorescience Dev project checked on 4 October 2026. The figures are illustrative project results, not a claim about current production performance.

This example shows how the same data becomes a clear report without losing its measurement context. It uses the Colorescience Dev project checked on 4 October 2026. The figures are illustrative project results, not a claim about current production performance.

## The question

“How often is Colorescience mentioned on each platform, and did its domain's citation share change between the selected completed runs?”

## 1. Fix the scope

Discover the project, then use `list_runs` to choose the runs completed on 19 and 24 September 2026. The latest run returned both readiness flags as true. Use its ID for current measurements and retain both IDs for the change comparison. Public request examples use placeholders so another account does not try to reuse this project's identifiers.

## 2. Check measurement quality

For the 24 September run, the domain-level stability response returned:

| Field | Observed value |
| - | - |
| `rank_corr_ok` | `true` |
| `sufficiency_ok` | `true` |
| `is_converged` | `true` |
| `structural_snr` | 2.3944 |
| `structural_snr_grain` | `domain` |
| `target_sufficiency_ratio` | 1.00 |

Report this as a converged domain-level measurement. It does not establish that every comparison will be significant or that every possible slice is equally reliable.

## 3. Read brand mentions by platform

Call `get_mention_score` with `metric: coverage`, `aggregation_level: entity`, `group_by: ["platform"]`, and `subject: {"kind":"specified","entities_or_bundles":["Colorescience"]}`.

| Platform | Mention coverage | Returned 95% confidence interval | Responses |
| - | -: | -: | -: |
| Gemini (`VERTEX`) | 39.17% | 33.21%–45.47% | 240 |
| ChatGPT Search (`OPENAI`) | 11.25% | 7.85%–15.87% | 240 |

These rows report the exact Colorescience entity, not the sum of its product aliases. They describe different platform conditions. The observed percentages alone are not a paired significance test, and the paired contrast tool in this edition does not test mention coverage.

## 4. Compare citation share across runs

Use `get_paired_contrast` for the bare domain `colorescience.com`, with the earlier run as A and the later run as B.

| Result | Value |
| - | -: |
| Run A citation share | 3.35% |
| Run B citation share | 2.67% |
| Difference, B minus A | −0.68 percentage points |
| 95% confidence interval on the difference | −1.94 to +0.26 percentage points |
| `significant` | `false` |

The interval crosses zero. A useful report says: “Observed citation share was lower in the later measurement, but this comparison did not establish a clear direction of change.” It should not say “visibility definitely fell” or “there was no change.”

## 5. Write the handoff

“Colorescience's selected domain-level measurement passed the convergence checks. Its measured mention coverage differed across the two platforms, so review their answer evidence separately. The selected citation-share comparison did not establish a clear run-to-run direction. Keep monitoring comparable runs and investigate the supporting sources before prioritizing a response to the observed decrease.”

This report distinguishes a measured fact, its uncertainty, and a proposed next step. [`worked-example.json`](/examples/worked-example.json) is a selected-field JSON excerpt of the observed responses, with account and run identifiers removed. Internal diagnostic fields are omitted.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.