AI visibility research
How AI crawls, ranks, and cites web content — sourced from provider documentation and ongoing observation. Updated as the AI ecosystem changes.
Every article cites its sources at the bottom · Methodology v2.0 (updated 2026-04-30) · RSS
Schema.org types Google AI Overviews actually surfaces — and the ones it ignores
2026-04-12AI Overviews lean on a specific subset of Schema.org markup to identify entities, products, and answers. Adding the right types is one of the cheapest visibility moves available.
PerplexityBot vs PerplexityUser: the two-bot model and what it means for opt-out
2026-03-05Perplexity operates two crawlers with different obligations under their stated policies. PerplexityBot respects robots.txt; PerplexityUser is treated as a user-initiated browser fetch and is not bound by it.
Google-Extended: opting out of Bard / Gemini training without losing Google search
2026-02-10Google introduced a separate user-agent token, Google-Extended, that controls only AI-product training. Blocking it does not affect Search ranking. Most publishers don't realise that Googlebot ≠ Google-Extended.
OpenAI's three web crawlers: GPTBot, OAI-SearchBot, and ChatGPT-User
2026-01-15OpenAI now operates three distinct user agents, each with different purposes and different opt-out implications. Blocking GPTBot doesn't keep you out of ChatGPT search results — and vice versa.