AI models are shortening the list of brands they recommend. That narrowing can make even category leaders invisible to users asking chat assistants for advice — and visibility in search engines no longer guarantees you’ll be on those shortlists.

A new analysis from Fractl, reported by Search Engine Land, compared brands’ traditional SEO footprints with how often large language models (LLMs) reference them. The headline: organic authority still matters, but a brand’s presence in third‑party content and the way models categorize it are increasingly decisive for AI recall.

What the study found, in plain terms

Most brands behave as SEOs expect: stronger search authority generally tracks with stronger AI visibility. But the interesting signals come from the exceptions.

  • About 5% of brands (471) had high domain ratings, deep keyword portfolios and millions of organic visits yet were underexposed in LLM outputs.
  • About 4% (377) were AI overperformers — the models referenced them much more often than their traditional search metrics would predict.
  • Roughly 9% of brands’ AI visibility aligned closely with how often they appear in third‑party content such as roundups, reviews and comparison pieces.

Visibility also fragmented across models: only 11% of brands in the dataset appeared across all three LLMs Fractl tested, while 77% were cited by just one. That means a single blended “AI visibility” number can hide important model‑specific differences.

Which brands win — and why this challenges SEO orthodoxy

LLMs tend to favor a compact set of default answers in each category. In travel, Booking.com (285 mentions), Airbnb (227) and Expedia (215) together accounted for roughly one in five travel recommendations. In HealthTech, Teladoc (275) led Amwell (220). In wellness, brands such as Peloton, Headspace, Calm, Whoop and Oura all registered significant mentions rather than a single dominant name.

Crucially, those favored names are not always the brands with the largest organic footprints. Legacy incumbents often under‑index. Examples in the dataset include legacy insurance carriers (Aetna, Cigna, Humana, Liberty Mutual) and major retail brands (Sephora, Samsung, Whirlpool) being displaced by digital‑native or direct‑to‑consumer names like Lemonade, Root, Patagonia and Allbirds.

Why? The models appear to draw heavily on the corpus of independent coverage that reiterates a brand’s role in a category: comparison articles, listicles and expert commentary. Owned content and technical SEO still matter, but corroboration from external sources shapes a model’s implicit category associations more strongly.

What marketers should measure and change now

The tactical implications are concrete: if AI recall matters for your funnel, add new metrics and new collaborators.

  • Track AI visibility by model and prompt category. Claude, Gemini and ChatGPT show different patterns: Claudelike fingerprints for SaaS and insurance, Gemini for travel and healthcare, ChatGPT the most consensus‑driven. Measure each model separately and map visibility by intent.
  • Audit third‑party coverage and category signals. Low recall often reflects miscategorization. Strengthen signals where models encounter them: analyst reports, industry press, buyer’s guides, partner pages and review sites that explicitly tie your brand to the category.
  • Pursue comparative placements and digital PR. Overperformers in the dataset repeatedly appear in other people’s content. Earn citations in roundups, expert lists and comparisons the models likely ingested.
  • Prioritize sectors where models are reshaping consideration sets. The analysis shows insurance and education have been notably re‑sorted: digital natives and credentialed providers over‑index. If you operate there, close the AI gap sooner rather than later.
  • Redefine your competitive set. The AI consideration set is smaller than the SERP competitive set. If you’re not among the five to ten names a model recalls, high Google rankings may not translate into AI referrals.

These moves push SEO teams to work closely with PR, analyst relations and partnerships. The objective is shaping the external narrative that models consume, not only optimizing owned pages.

What to watch next (and what to do first)

Start with a simple audit: which models mention your brand, for which prompts, and which external sources are repeatedly cited when the model recommends a brand. That map tells you whether you need clearer category signals, more third‑party corroboration, or model‑specific outreach.

Expect vendors to build model‑specific visibility tools and dashboards. Meanwhile, prioritize the low‑effort wins: secure placements in comparison content, brief analysts and partners on category framing, and instrument visibility by model so you can act where it matters.

AI visibility is part measurement problem, part distribution problem. Brands that align SEO, PR and category‑focused content — and treat third‑party corroboration as a core KPI — will be most likely to appear in the shortlists these models produce.