How does ChatGPT pick the sources it cites?
What OpenAI documents about ChatGPT search, which crawler it uses, how to verify access, and what my own Observatory and benchmark data show.
Guides
How AI visibility works, drawing on official documentation, research and my own data.
Thematic guides
How does Google pick the sources in AI Overviews and AI Mode?
What Google documents about its generative features, what the site controls, how to verify access, and what my Google AI Mode data in the Observatory shows against ChatGPT in the same cut.
How does Copilot pick the sources it cites?
What Microsoft documents about Copilot's web search, what Bing Webmaster Tools reports back to a site owner, and what the AI Performance report for aurametrics.io shows across its 91 grounding queries.
Why does a competitor appear more often than my brand in AI answers?
What engines document about how they pick sources, what research supports and under which condition, and how to measure the associated signals in a specific case.
What replaces rank tracking when an AI gives the answer?
Why position stops applying once the engine rewrites the query, what gets measured instead, and how to start measuring it in your own case.
Which external sites shape what AI cites about a brand?
What engines document about the sources they show, how citations split between a few domains and a very long tail, and how to measure which sites travel with a category's answers.
How can I tell whether AI names competitors instead of my brand?
Seven steps to measure whether ChatGPT, Gemini and Claude name the competition when they answer about a category: what to look at in each step, which module covers it, and three published measurements showing how the leader changes by market, by language and against its runner-up.