Crawler access and policy gates
Robots, headers, and access controls determine whether AI crawlers can reach the pages at all.
Three things have to be true before your business can be quoted in an AI answer, and each one fails silently:
Access. A rule in one text file can keep every AI crawler off your site, and nothing on the site looks wrong.
Understanding. Without machine-readable identity, you are text on a page rather than a named business with an address and hours.
Citation-readiness. A page with no stable URL, no date and no author is hard to attribute, so it is easier to cite a competitor.
This scan reads your public configuration and reports what it actually observed. It does not measure how often ChatGPT, Perplexity, Claude or Google AI Overviews mention you — nothing available from outside your own analytics can.
Illustrative sample UI
Example dashboard structure, not a real result.
Sample only. Your scan reports what was actually observed on your public pages and marks anything unavailable as untested.
The interface below follows the same path: access signals first, then rendering and content, then the evidence that supports citation and entity matching.
An anonymised report from a real scan of a single-location dental practice. Nothing here is invented for the sake of the illustration.
Grade C+ · Partly configured
example-dental.com · 6 groups scored, 2 checks not testable
Top fix in that report
Add LocalBusiness structured data. Worth 8 points, a configuration change, affects every assistant that reads schema. Typically visible to crawlers within days of the next crawl.
The landing page keeps the domain form first, then shows the exact infrastructure path the scan audits.
Crawler access and policy gates
Robots, headers, and access controls determine whether AI crawlers can reach the pages at all.
Rendering, content, and freshness
The scanner checks whether the page can actually be rendered, parsed, and cited with intact content.
Identity and citation evidence
Entity signals, stable references, and corroboration determine whether an assistant can confidently name you.
Every finding says what was observed, on which URL, and what it costs you.
Scan again after a fix and the report shows what moved, what is new, and what is still open.
A check that could not run is reported as untested, never scored as a zero.
Powered by The Infrastructure Agent
Each line above is marked complete only after that check has returned an answer. A check that fails is reported as untested and the rest keep running.
—
Part of this scan could not be completed
Untested checks are excluded from the score rather than counted as failures, so the percentage above is out of what was actually measured.
Ranked by the points available against the work required. Propagation times are how long a change typically takes to become visible to crawlers, not a promise about any assistant's output.
Every check, what was observed, and what it affects.
Not part of your score
Each competitor is scanned with the same model, one at a time. Categories where a check could not run on every site are left out of the comparison rather than counted as a zero.
This runs a full scan of each domain you enter, so it counts against the same hourly limits as your own scan.
We run a set of market-specific queries against the Brave Search index — one independent web index — and add a prioritized 90-day action plan.
Free — takes about 30 seconds. This is not a measure of ChatGPT, Perplexity, Claude, Copilot or Google AI Overviews visibility.
Powered by The Infrastructure Agent · AI Search Consulting
We’ll query the Brave Search index for your business and send you the results with a prioritized action plan.
What happens to these details
Querying the Brave Search index for your business.
This is one query set against one independent web index. It usually takes under a minute, and there is nothing to report until it answers.