Matching the dashboard
Which report gives the numbers of which app screen, and the few cases where the API and the app count differently.
Each report counts answers by the same rules as a screen in the app, so its numbers match that screen:
| Report | Gives the numbers of |
|---|---|
| visibility, no dimension | the Overview page |
visibility, prompt dimension | the stats on the Prompts page |
| visibility, sentiment counts | the sentiment spread |
| brands, no dimension | the competitors leaderboard and the share of voice panel |
brands, day dimension | the visibility comparison chart |
citations, domain dimension | the domains table on the Citations page |
citations, url dimension | the URL table on the Citations page |
Use the same dates in both. The app's date picker and the API both use UTC days, and both include the first and the last day.
Archived prompts
Every list and report leaves out the answers, mentions, and citations of archived prompts, as every screen in the app does. To count them, send include_archived=true:
| App screen | API request |
|---|---|
| Results, with Show archived prompts turned on | /v1/answers?include_archived=true |
| The page of an archived prompt | the same list or report with prompt_id=<id>&include_archived=true |
The app opens the page of an archived prompt without any extra setting. The API needs include_archived=true when prompt_id names an archived prompt. Without it, the request gets status 400 with the error code filter_value_invalid. This way, a filter never returns 0 rows without telling you why.
Percentages
The API sends rates as fractions, for example visibility: 0.4199. The app multiplies them by 100 and shows 42%.
Visibility and share of voice count different answers
visibilityandmentionscount an answer when it ranks your brand or names it in the text.share_of_voiceandrankedcount only the answers that rank the brand in their list.
The app makes the same split: the visibility comparison chart counts answers that rank or name a brand, while the share of voice panel and the leaderboard count only ranked answers.
Where a list and a report count differently
- Answers with no text. Reports count every completed answer, as the dashboard does.
/v1/answerslists only answers that have text, as the Results page does. So the rare completed answer with no text is counted in a report but does not appear in the list. - Prompts with no answers. The Prompts page shows a prompt that has no answers in the date range as a row of zeros. The visibility report has no row for it.
/v1/promptslists every prompt. - Google searches with no AI Overview count only in the rows of a report split by engine. See Engines.
- Answer dates. Lists and reports filter on
answered_at, the time the answer was completed. The Results page filters on the time the answer was started. The 2 differ only for an answer that finished on a later day than it started.