Going into this examine, I suspected that domain-level hyperlink metrics wouldn’t be predictor for mentions. I used to be anticipating Google AI Overviews to point out extra of a correlation since they might use a few of the conventional search alerts, whereas the opposite techniques might not. For conventional search, normally the page-level metrics are extra predictive of rankings than domain-level ones.
I seemed on the prime 50 web sites talked about in Ahrefs Model Radar for Google AI Overviews, ChatGPT, and Perplexity. That is throughout ~76.7M AI Overviews, 957k ChatGPT prompts, and 953.5k Perplexity prompts for the month of June 2025.
I in contrast the web site mentions to their Ahrefs Rank (AR) in Ahrefs. Ahrefs Rank (AR) ranks all of the web sites in our database so as, by the dimensions and high quality of their adopted referring domains.
Sidenote.
Model Radar isn’t simply one other LLM visibility monitor, we observe a considerable amount of queries throughout all of those techniques and you may question for any product, service, or model and examine vs rivals. It’s extra like Website Explorer than it’s Rank Tracker. Plus we now have the online visibility index with the intention to see the way you’re talked about on-line and the search demand index to see how fashionable you might be in searches.
Right here’s what Model Radar appears to be like like.
Let’s dig in.
Perplexity reveals a reasonable correlation between the mentions of its most-cited domains and their Ahrefs Rank, Google AI Overviews a really weak correlation, and ChatGPT no correlation. I used to be anticipating Google to have a stronger relationship than Perplexity, so the outcomes stunned me.
This correlation would probably change with a bigger dataset exterior of simply the highest 50. The highest 50 are all extraordinarily well-linked and cited web sites.
I’ll give the standard correlation doesn’t equal causation disclaimer.
Listed below are the Spearman rank correlations for point out share vs. Ahrefs Rank throughout the highest 50 domains in every system:
| AI Assistant | Spearman Rho | P-Worth | Correlation |
|---|---|---|---|
| Google AI Overviews | -0.12 | 0.47 | Very weak |
| ChatGPT | 0.01 | 0.95 | No relationship |
| Perplexity | -0.34 | 0.095 | Weak |
Google AI Overviews over-index on UGC websites like YouTube, Reddit, and Quora, in addition to encyclopedic content material like Wikipedia. We’ve seen this in just a few different research already.
These websites get further weight in mentions than their hyperlink profile predicts.
Google AI Overviews underweights on social media websites like LinkedIn, Instagram, and Fb. One other sample we’ve seen in different research.
We additionally see an analogous story with ChatGPT, the place they’re obese on issues like Wikipedia and information websites like Reuters, precisely like we noticed earlier than.
Perplexity additionally reveals the identical biases we’ve seen. YouTube and Wikipedia are obese, social media websites are underweight.
Closing ideas
I’m undecided that is sufficient of a pattern measurement to get measure. We really noticed larger correlations in Google AI Overviews in comparison with DR after we studied 75k websites than we noticed in correlations to regular Google rankings throughout 1 million key phrases, 0.326 vs 0.131.
We’ll run a much bigger examine past the highest 50, in addition to take a look at some extra web page degree metrics for extra insights.
