Answer engines
Answer engines do not behave as one. They search different indexes, cite different numbers of sources and know different brands from memory, so Vercatus keeps every engine’s figures apart and never folds them into one figure without showing the parts.
The public summary, with a real comparison of how engines answer one question, is on the Engines page.
Named, cited, top pick and read
Section titled “Named, cited, top pick and read”Within one answer, a brand can be:
- Named: the answer mentions it.
- Cited: the answer links one of the brand’s own pages as a source.
- Top pick: the answer’s first or only recommendation.
A page can also be read without being cited: the engine retrieved it, and the answer does not credit it. Read and cited are kept apart everywhere. A page an engine only read is never counted as a citation, but it is counted in the Sources lens, because the pages an engine reads shape whom it names. See Why.
How each engine is reached
Section titled “How each engine is reached”| Engine | How it is reached | The index it searches |
|---|---|---|
| ChatGPT | Its app, as a person sees it; its official interface with web search on when the app cannot be read | Bing, in part |
| Google AI Overviews | Observed on Google’s live results page; the Overview’s references are its citations | |
| Google AI Mode | Observed on Google’s AI Mode page; its references are its citations | |
| Gemini | Grounded on Google Search; a source counts only when the answer’s text rests on it | |
| Copilot Search (Bing) | Observed on Bing’s live results page; the summary’s references are its citations | Bing |
| Claude | With web search on; it cites the passages it quotes | Brave Search |
| Perplexity | Searches the live web for every answer | Its own index, not checked |
| Grok | With live web search on | Not checked |
| Le Chat | With web search on | Not checked |
| You.com | Its research interface, in a full run’s audit only; the poll and the Engine Briefs do not ask it | Not checked |
The index column is what the Search lens checks: where you and your competitors rank, in the index that engine reads, for the searches the engine ran. Google, Bing and Brave Search are checked. An engine whose index is not checked is shown as “not checked”, never as absent.
Engines that do not answer with their sources through any interface Vercatus can read (Meta AI, Siri and others) are not measured. For those, the crawler catalogue reports whether the site lets their crawlers in, which is the precondition, not the citation.
Google’s AI answers
Section titled “Google’s AI answers”AI Overviews, AI Mode and Bing’s Copilot Search cannot be asked; they are observed. For each question, Vercatus reads the live results page in the country and language of the question’s market. Google shows an Overview for some questions and not others. A results page without its AI answer is recorded as “no AI answer shown” and counts neither for nor against anyone: there was nothing to name.
Where engines show in the app
Section titled “Where engines show in the app”- Overview: the engine strip, your share and the leader on each engine. ChatGPT is marked “app” on the days its answers came from the app.
- A topic’s page: who is recommended, with a cell for each engine.
- Why: the Search lens per index, and the Memory lens per engine.
- Briefs: what each engine says about you.
- Plays: each play names the engines its evidence rests on, and its verdict treats those engines.
The engine picker in the top bar narrows the screens that read it to the engines you choose. The choice is kept in the address, so a link opens the same view. Each engine has one colour and mark wherever it appears.
Crawlers
Section titled “Crawlers”Each engine’s crawlers are separate from the engine. Whether they can read your site is in Crawlers & access, from the daily access watch; what each crawler is for is in Crawler tokens.