How to Let ChatGPT Read Your Site
OpenAI runs three crawlers and only one of them decides whether you appear in answers.
OpenAI runs three crawlers and only one of them decides whether you appear in answers.
AI Overviews live inside Search, so the control isn't the one most people reach for.
Two settings decide whether PerplexityBot gets in. Both take a minute to check.
Anthropic runs three crawlers. Blocking the training one costs you nothing.
There's no AI-only switch here: one crawler feeds both Copilot and Bing search.
Five crawlers, easy to confuse, and one of them controls your link previews.
One of Apple's two names isn't a crawler at all, which catches almost everyone out.
A three-line file decides where crawlers may go. One stale rule can hide your services.
Without a sitemap, pages nobody links to may never be visited at all.
A tag left on after launch keeps a perfectly good page out of every answer.
If your menu isn't made of real links, everything behind it stays unvisited.
A dead link ends the path and takes everything behind it along with it.
One line can tell assistants your page is a copy of something else.
Your words are there for visitors and missing from the file crawlers read.
No assistant can sign in or click “I accept”. Behind that, nothing exists.
Crawlers don't wait. Slow answers mean fewer pages read, less often.
The page opens, but the files it needs in order to display don't.