User agent
The text a browser or app sends to identify itself, from which device class, platform and crawler identity are inferred.
Updated August 14, 2026
The user agent is a string every request carries to describe the software making it. A redirect service parses it to fill the device, operating system and browser breakdowns, and to decide whether a visitor on a phone should be offered an app rather than a web page.
The same string is also how automated traffic announces itself. Search crawlers, messenger preview fetchers and monitoring tools all name themselves there, which is convenient and completely unverified: the header is set by whoever sends the request, so any script can claim to be a search engine. That is why traffic rules treat a declared crawler as confirmed only when its address range agrees with the name, and match text patterns as a hint rather than as identity.
Two habits follow. Do not build durable logic on exact version numbers — modern browsers freeze and simplify the string on purpose. And when a device breakdown shows an implausible spike, check the user agents behind it before rewriting the campaign; a single misbehaving tool explains most of them.