Empirank

Original Study · 16 Aug 2026

Backlinks Didn’t Explain AI Mentions Once Reviews Were Taken Into Account

We tested 500 US plumbing businesses. Better-linked companies appeared more often at first, but referring-domain count no longer explained mentions after reviews, market, and query differences were considered.

Study sample
500 businesses
Measured outcome
Business recommendations
AI platform
OpenAI

What this means for marketers

Keep building relevant links for rankings, referral traffic, and brand discovery. Don’t sell raw link count as a direct AEO tactic.

  • Don’t buy links on the promise of guaranteed AI mentions. This study found no evidence for that claim.
  • Earn relevant links for outcomes they can support: rankings, referral traffic, discovery, and brand awareness.
  • Keep your review program running. The review pattern remained after we added links to the model.
  • Judge links by audience and relevance, not just domain count or an authority score.
  • Measure the AI platform your customers actually use. Results from one setup may not transfer to another.

What we tested

Do plumbing websites with more referring root domains get mentioned more often in search-enabled OpenAI recommendations after review count is taken into account? The study covered 500 businesses.

What the result does not prove

This was an observational study of one industry and one backlink index. It cannot tell us how links behave across every market, page, or AI platform.

  • This was a follow-up using the same businesses and AI answers as Study 1, not an independent replication.
  • We collected backlink data after the AI answers, so the two measurements aren’t perfectly time-aligned.
  • DataForSEO is one backlink index and may miss links that another provider sees.
  • Domain-level metrics may not reflect the authority of a specific local landing page.