Where ChatGPT, Perplexity, and Google AI actually get their answers

Ask ChatGPT and Perplexity the same question and you often get the same answer built from different sources. One leans on an encyclopedia. The other leans on a forum. That split is not noise. It is the most useful thing to understand about AI search.

Profound analyzed 680 million citations across ChatGPT, Google AI Overviews, and Perplexity. The patterns are stark.

The engines have different taste

Inside ChatGPT's top ten sources, Wikipedia alone takes nearly half. Inside Perplexity's, Reddit takes almost half. Google spreads its trust wider, across forums, video, and professional networks. Three engines, three philosophies.

What this means for you

You cannot win AI visibility with one move. Being strong in Google's index helps you in AI Overviews, because that feature runs on Google's ranking systems. It does much less for Perplexity, which is busy reading Reddit.

A few things follow directly:

• Earned presence on third-party sources matters. AI often cites Reddit, Wikipedia, G2, and industry sites more than your own domain. Being mentioned there, authentically, shapes how AI describes you.
• Community is a distribution channel now. Honest participation in the right Reddit threads and review sites is no longer just brand work. It is AI-visibility work.
• Your own site still has to be excellent. Citations skew heavily commercial: .com domains account for over 80% of ChatGPT's citations. Your best content is still your best asset.

A caveat worth keeping

This is a snapshot, not a law. The data covers a window in 2024 and 2025, and the engines change their sourcing constantly. Treat the specific percentages as a weather report, not a constant. The durable lesson is the one that will outlast the numbers: each model trusts a different mix, so you have to watch each model.

Which raises the obvious question. How do you watch them?

Run your free audit → Back to Academy