Artificial intelligence is transforming online search. To succeed online, you need to know what drives AI search visibility. A recent study looked at the link between domain-level link metrics and mentions in AI assistants like Google AI Overviews, ChatGPT, and Perplexity. The results are surprising. They raise questions about how much traditional SEO indicators still matter in this new era.

Shifting Search Paradigms: AI Search Visibility

Traditionally, domain-level link metrics – such as the number and quality of referring domains to a website – have shaped how prominently a site appears in search results. These signals reflect a site’s authority and trustworthiness, especially to search engines like Google.

But AI-powered search tools now generate their own answers from across the web. So the weight of traditional link signals may be fading. AI systems pull information from many sources. They don't always favour the signals that drive conventional search rankings.

To explore this, a study reviewed the top 50 most-cited websites across major AI tools in June 2025. The analysis used data from tens of millions of AI-generated responses to uncover patterns in AI search visibility.  It compared those mentions against each site's overall link authority to see if a clear correlation still held.

Key Findings: Weak to Non-Existent Correlations

Here’s what the analysis found. It compared how often each site was mentioned with its position in global domain authority rankings:

AI Tool

Spearman Correlation (Rho)

P-Value

Interpretation

Google AI Overviews

-0.12

0.47

Very weak relationship

ChatGPT

0.01

0.95

No relationship

Perplexity

-0.34

0.095

Weak/moderate link

Perplexity showed the strongest correlation between domain authority and mention frequency, though it was still weak. ChatGPT showed almost no connection. Google’s AI Overviews comes from a search-focused company, yet it had only a very slight relationship.

These findings challenge a common assumption. Many believe AI-generated content tools lean heavily on domain strength when picking sources. In reality, AI systems seem to prioritise other signals. They may favour relevance, content type, or recency over link-based authority.

A Closer Look at Each AI Assistant: AI Search Visibility

Google AI Overviews: Prioritising UGC and Encyclopaedic Content

Google’s AI Overview feature clearly prefers user-generated content and encyclopaedic-style sources. It cites sites like YouTube, Reddit, Quora, and Wikipedia often. It cites them far more than their domain link metrics would suggest.

Platforms long seen as highly authoritative get fewer mentions. That includes Facebook, LinkedIn, and Instagram, all strong on link profiles. This may signal a shift in Google’s AI priorities. It favours depth and community insight over raw link metrics.

ChatGPT: Following Established Media and Reference Sources

OpenAI’s ChatGPT also leans towards factual, encyclopaedic content. Wikipedia and major news sites like Reuters feature prominently. Yet even with their presence, there’s no meaningful link between how often a domain is cited and its domain-level link authority. ChatGPT’s retrieval model seems to value topical relevance and trusted information over traditional SEO factors.

Perplexity: Slight Nod Towards Authority, But Still Content-Led

Perplexity is a newer AI-powered search engine. It's the only platform to show even a modest correlation between mentions and domain authority. Still, that link is fairly weak. So Perplexity also prioritises the quality and context of content over pure link metrics.

Like the other tools, Perplexity leans towards Wikipedia and YouTube. It pulls less from social media or high-DR commercial sites. This fits a broader trend. AI systems appear to favour source clarity, information density, and structure.

The Role of Page-Level Metrics: AI Search Visibility

One reason for the weak correlation may be the focus on domain-level data instead of page-level performance. In traditional SEO, individual pages with targeted, high-quality content often rank better. So AI search visibility may depend more on page relevance than overall domain strength.

An earlier study of 75,000 sites found stronger correlations between AI Overviews and domain ratings. AI search visibility also outperformed traditional rankings across one million Google keywords. This suggests AI-powered tools may weigh different factors altogether. Traditional link-based SEO metrics may not yet capture them.

What Does This Mean for Content Strategy?

To appear in AI-generated answers, brands and content creators may need to adjust their strategies. Here are a few practical takeaways:

  • Prioritise clear, structured information: AI assistants seem to favour encyclopaedic, informative content, especially on platforms like Wikipedia or Quora.
  • use user-generated content: AI tools often cite forums, Q&A sites, and long-form discussions, particularly Google.
  • Don’t rely solely on backlinks: A strong link profile still helps organic search. But it may not do much in AI environments.
  • Optimise for topical depth and authority: Thorough, well-researched pages on niche topics may beat generic, highly-linked content.
  • Stay updated: The algorithms behind these AI assistants evolve quickly. A strategy that works today might be outdated tomorrow.

Final Thoughts: A New SEO Frontier?

AI-assisted search is still in its early stages. The patterns we see here may shift as these technologies mature. One thing is clear, though. The traditional SEO playbook may not fully apply in AI-driven contexts.

Do backlinks still influence AI? Less than they used to. AI search visibility now depends on content depth, quality, and contextual relevance. As AI reshapes search, SEOs, marketers, and content creators must adapt to stay ahead. AI search visibility now demands relevance and agility, in a space where the rules shift in real time.

Future research will dig into larger datasets and add page-level comparisons. That should give us a clearer picture of what really drives prominence in AI-generated results. Until then, focus on relevance, clarity, and user value. It may be the best way to stay ahead.