How Geordy measures AI crawler activity.

What we observe

We observe requests to the files Geordy serves: files on files.geordy.ai and on our customers' own Geordy subdomains. We do not observe requests to customers' HTML pages.

A request to one of our files is a request, nothing more. It is not evidence that a model ingested the file, that an answer engine recommended a site, or that anything was cited.

How crawlers are identified

A crawler is identified by the user agent string it sends. We do not verify requests by IP address or network checks, so a request that claims to be a given crawler may not be one.

What we publish

We publish shares, ranks and trends across sites. We never publish raw request counts, and we never publish any customer's traffic.

No figure is shown until a minimum number of independent sites support it. Every figure carries one of four evidence levels: Established pattern, Emerging pattern, Limited observations, or Not measured.

Documented vs observed

Where a provider documents how its crawler uses a format, we cite the provider's documentation and say so. Where we state that something is observed, it means we saw it in our own logs, under the rules on this page. The two are never mixed: a format that is written for a reader is not, for that reason, read by it.

Status today

Public measurements are not published yet. They will appear on crawler pages and on format pages, such as the format library, once enough independent sites are covered. Until then those pages say who a format is written for, and make no claim about who requests it.