The free audit asks Perplexity a handful of questions a buyer of yours might ask, and shows whether it cites you. It does not ask Claude: that is measured in the product, on Starter, Growth and Scale.
Anthropic says that when you ask about something that benefits from current information, Claude uses a search tool to ground its answer in the live web, and its responses include citations.
In Anthropic's API, Claude searches when a request depends on information that is current, changing or outside its training data, and answers directly otherwise.
Citations are always on for API web search. Each carries the URL, the title and up to 150 characters of the cited text, and Anthropic requires them to be shown to end users.
Claude-SearchBot indexes pages for Claude's search results. Anthropic says blocking it may reduce your site's visibility in them.
Claude-User retrieves a page for a user's question, and ClaudeBot collects training data. Anthropic says all three honour robots.txt.
How we ask
We put each tracked question to Claude through DataForSEO's LLM Responses API, using the claude-haiku-4-5 model with web search on, and read the sources the answer cites. That is the engine's API, not a signed-in session in the Claude app, so an answer can differ from the one a person sees.
How often
Every 7 days per question, spread across that window so a domain's checks do not all land on one day. On trial accounts, scheduled checks pause after 14 days without a sign-in and resume on the next visit.
Each check records
| Plan | Tracked questions per domain | Checks a month, all engines |
|---|---|---|
| Starter | 50 | 500 |
| Growth | 100 | 1,200 |
| Scale | 100 | 4,000 |
Past the monthly budget, the questions checked longest ago go first and the rest wait for the next cycle.
Vendor sources checked 2026-09-24