Visibility history
The history keeps every test run and lets you see whether your visibility is improving, holding or falling. It is available both in aggregate, on the per-query-type cards, and in detail inside each query.
What gets recorded#
Every time a query is tested, AgentFi sends it to ChatGPT, Perplexity, Gemini and Claude, and records whether your site appears in each answer. That record is not wiped when you test again: it accumulates.
Per-engine history#
In the detail of a query, each engine gets its own row. The coloured dots are individual runs, oldest on the left, most recent on the right.
The colours#
| Outcome | What it means |
|---|---|
| Found in answer | The engine cited or named your site. When a citation position is available it reads Found at #N: lower numbers mean the engine placed your site higher in its list of sources. |
| Not mentioned | The engine answered the query, but neither cited nor described your site. This is the metric with the most room for improvement. |
| Engine error | The AI engine did not return a usable answer: provider outage, timeout or content filter. It counts neither as found nor as not mentioned, so it does not penalize your percentage. |
The daily squares on the cards#
On the Visibility by query type cards, each square represents one day of history for one engine. The colour reflects that day's success rate for that engine, following the five-band legend.
Squares only appear on days when tests ran. With weekly auto-tests you get one square per week; at a daily cadence the grid fills in and trends become far easier to read.
Strength labels#
| Label | How to read it |
|---|---|
| Strong | The engines find you consistently. |
| Moderate | They find you sometimes. There is clear headroom. |
| Needs work | They rarely find you. Review optimization coverage and your custom rules. |
| No data | Not enough tests yet. |
How to use the history#
- Set a baseline: run every query once, before changing anything.
- Make one change at a time — optimize a group of pages, adjust the custom rules, publish new content.
- Wait for the next scheduled run and compare the discovery visibility card.
- Look at the queries that moved from Not mentioned to Found in answer and work out what they have in common.
AI engines are not deterministic: the same query can return different results in two consecutive runs. Never draw conclusions from a single test; look at the trend across several runs.
Reputation analysis#
Beyond measuring whether you appear, AgentFi can analyse how your brand is talked about. Reputation analysis generates its own queries, runs them across all four engines and analyses the sentiment of the answers.
Reputation analysis spends tokens: on top of the generated queries, the system creates follow-up queries while studying the sources it finds, and each one costs tokens.