Amzn-SearchBot

In server logs the bot identifies itself with a user agent that contains the token Amzn-SearchBot, for example Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Amzn-SearchBot/0.1) Chrome/W.X.Y.Z Safari/537.36. Amazon says the visits feed search experiences inside its own products and services, and names Alexa as one of them. Permitting the crawler makes your content eligible to appear in those experiences. Amazon separates this bot from model training in its documentation, which states that Amzn-SearchBot does not crawl content for generative AI model training.

Blocking the token removes your pages from eligibility for Amazon search experiences such as Alexa, since the operator ties eligibility to permitting the crawler. One robots.txt detail deserves attention. If your robots.txt does not mention Amzn-SearchBot but allows other search bots, the crawler follows the directives given to those other search bots. To exclude it while allowing others, name the token explicitly. Amazon documents the same robots.txt handling as Amazonbot: a host-level robots.txt, a 30-day cache, support for rel=nofollow and the meta noarchive, noindex and none values, no crawl-delay support, and about 24 hours for setting changes to take effect. Each user agent setting is independent of the others.

Verification relies on the published IP list alone. The page at developer.amazon.com/amazonbot/searchbot-ip-addresses/ is HTML with an embedded JSON blob whose prefixes array uses the field name ip_prefix, with each value carrying a /32 CIDR suffix. No reverse DNS suffix is documented, so a hostname check cannot confirm a request. Amazon does not document whether the bot executes JavaScript.