New research from Yext (NYSE: YEXT) reveals the inner workings of the AI engines increasingly used by consumers in place of traditional search. Though this usage continues to grow, the rules of engagement for companies seeking AI search visibility remain unclear. And according to Yext, AI SEO may be harder than it looks.

Among other things, the report reveals that AI search visibility varies by model, sector, and industry. By examining 17.2 million distinct AI citations across Claude, Gemini, Perplexity, and SearchGPT, Yext found that there was a wide variance between each engine in terms of where they source info.

As a result, a given marketer can apply rigor to successfully gain visibility in one engine… only to be underexposed in another. This means that there isn’t a firm playbook like there is in SEO – though that’s never been terribly concrete either. It also means marketers have new rules to learn… and lots of them.

This sends a daunting message to marketers: Not only do they have to learn a new language, but they also have to master several of its dialects. On the bright side, this offers an opportunity for ambitious marketers to gain an edge while others fall behind, due simply to the magnitude of new tactics to learn.

“Every AI model cites differently, and the variation within sectors is larger than even we expected,” Yext Research Senior Director, Adam Abernathy, told Localogy Insider. “A restaurant might look great in Gemini but could be invisible in ChatGPT, because one is reading your website and the other is reading your Yelp reviews.”

Yext Reports High Consumer Trust in AI

Boiling it Down

Going deeper into the results and boiling things down for Localogy Insider readers, here are some concrete takeaways we extracted from the report.

• Model Machinations: Claude cites reviews and social content at 2–4x higher rates than other top AI models across all seven sectors studied.
• Vertical Variations: Yext found large variances in how various vertical categories performed across models.
  • In retail, general merchandise sees 64.99% of citations from third-party listings, highlighting heavy aggregator reliance.
  • In hospitality, SearchGPT cites official hotel websites 38.1% of the time, roughly double competing models (16.7–22.4%).
  • In food & beverage, Claude sources 24.35% of citations from reviews/social, nearly 10x Gemini’s rate.
  • In financial services, banking & lending shows 58.52% reliance on directories/comparison platforms.
  • In healthcare, specialized physicians see 58.39% directory reliance, with the narrowest model divergence of any sector.
  • In non-profits, 25.62% of entities have “no control” citations, meaning visibility is largely shaped by independent media and community sources.

Yext: The Path to Purchase is Longer than Ever

It Depends

Boiling all of this down into an actionable takeaway, these results indicate that there’s no one-size-fits-all AI search strategy. When marketers are looking to determine which sources and citations have the most impact – from websites to listings to review sites – the answer depends on the model and the vertical.

Another practical takeaway is that marketers who are wading into the waters of AI search need to adjust their expectations about the time required to master it. And they need to prepare themselves to reach a level of mastery that’s likely greater than expected. If so, they can gain an edge in the AI search era.

“Businesses that treat AI search as a singular channel are optimizing for an average that no one actually experiences,” said Abernathy.

Header image credit: Solen Feyissa on Unsplash