ChatGPT Citations Track Google Rankings More Than Overlap Implies

What happened

A common reading of recent AI-search research is that ChatGPT barely overlaps with Google, so Google rankings do not matter much for being cited there. A small study of sites hit by a Google algorithm update cuts against that reading.

In February 2026, Lily Ray examined 11 sites that lost Google rankings in an update and tracked what happened to their AI citations. ChatGPT citations fell 27.8%. That was steeper than the 23.8% drop in Google's own AI Mode. Perplexity citations moved only 2.9%. 2

On its face, ChatGPT was more sensitive to Google ranking losses than Google's own AI product. That is hard to square with the overlap data.

The overlap numbers that seemed to say otherwise

Several large datasets suggest ChatGPT's citations sit mostly outside Google's top results. Ahrefs analyzed 15,000 prompts and found only 12% of URLs cited by assistants such as ChatGPT, Gemini, and Copilot rank in Google's top 10 for the original query. For ChatGPT specifically, the figure was roughly 6 to 8%. 1 Another breakdown puts ChatGPT's top-10 overlap near 6.5%. 3 A separate tally gives ChatGPT 8%. 4

Semrush came at it from the other side. It found the pages ChatGPT cites rank in position 21 or lower almost 90% of the time. 1

The broader citation landscape looks fragmented too:

• Google AI Overviews' overlap with the organic top 10 reportedly fell from about 76% to between 17% and 38%. 4
• Only about 11% of cited domains are shared across engines. 4
• Each engine leans on a different kind of source. ChatGPT favors Wikipedia and authoritative media, Perplexity favors Reddit, and Google's AI surfaces mix YouTube, Reddit, Wikipedia, and over-indexed .gov pages. 1
• Ahrefs found that 67% of ChatGPT's top 1,000 cited pages are Wikipedia entries, brand homepages, or app store listings. 3

Where the explanations diverge

The sources disagree on what actually powers ChatGPT's retrieval.

• Bing: Two of them say ChatGPT's search runs on Bing. They cite Seer Interactive's controlled study, which found 87% of SearchGPT citations matched Bing's top results versus 56% for Google's. 34
• Google: The analysis built on Ray's study argues instead that ChatGPT appears to scrape Google during live searches. That would explain why Google ranking losses hit it so hard. 2

The sources also differ on Perplexity's plumbing and how closely it tracks Google.

• Plumbing: One describes Perplexity as using the Brave Search API, which would insulate it from Google ranking shifts. 2 Another describes it as running its own Sonar pipeline. 3
• Overlap with Google's top 10: Ahrefs puts it around 28.6%. 1 A comparison table elsewhere lists 82%. 3 These figures likely reflect different methods and should not be reconciled casually.

A reading that fits both facts

My read is that the low-overlap statistic is being asked to prove something it cannot. A 6–8% match with Google's top 10 says ChatGPT rarely cites the same exact URLs that rank highest for the same query. It says nothing about whether the systems feeding ChatGPT reward the same underlying signals.

The Semrush finding points in this direction. If most ChatGPT citations come from pages ranking 21st or lower, then a site that falls out of positions 15 through 50 still loses citation opportunities. A top-10 overlap metric would never register that loss.

The Bing-versus-Google dispute may also matter less than it appears. If a Google update demotes a site for quality or authority reasons, other search indexes may judge those pages similarly. A steep ChatGPT decline would then show up even if ChatGPT's retrieval runs through Bing. That is speculation, but it explains the 27.8% drop without requiring the contested claim that ChatGPT scrapes Google directly.

Perplexity's near-flat result is the more telling contrast. Whatever the exact backend, it appears to sit further from Google's ranking signals than ChatGPT does.

Caveats

Eleven sites is a thin sample. The study captures one update, not a durable relationship. The sources also measure overlap in incompatible ways, which is why Perplexity can look like either a 28% or an 82% match depending on the dataset.

What it means for SEO

The practical advice converges even where the explanations do not:

• Track citations per engine. A single AI visibility number hides very different behavior across platforms. 2
• Don't assume one strategy transfers. The cross-engine overlap is low enough that optimizing for one answer engine will not reliably carry over to others. 4
• Keep investing in organic search. Low overlap is not the same as independence. For ChatGPT at least, losing organic standing appears to cost real AI visibility, even when the cited URLs rarely sit on page one.