How AI engines choose which brands to name

Two inputs: what the model learned, and what it retrieves

An AI answer draws on the model’s training data and, for AI search engines, on pages retrieved live for the question. Your brand has to be present in both.

Training data shapes which brands a model “knows” and how it describes them. Live retrieval decides which pages it reads for a specific question and, in AI search products, which ones it cites. ChatGPT search, Perplexity, Gemini and Google AI Overviews all retrieve pages at answer time. Google describes a “query fan-out” technique, “issuing multiple related searches across subtopics and data sources”, to find supporting pages ^[5].

What the original GEO research measured

Making a page more quotable and better sourced raised its visibility in generated answers. Classic keyword tactics did not.

The term “Generative Engine Optimization” comes from a 2023 paper by a team including Princeton University researchers, published at KDD 2024 ^[1]. They tested content changes on a benchmark of 10,000 queries and report that GEO methods “can boost visibility by up to 40%” in generative engine responses.

• What worked best: adding quotations, adding statistics, and citing sources. The best methods improved on the baseline by 41% and 28% on the paper’s two visibility measures.
• What did not: “simple methods like Keyword Stuffing traditionally used in SEO don’t perform well.”
• Who gained most: lower-ranked pages. Citing sources raised visibility by 115.1% for pages ranked fifth in search results, while it lowered visibility for pages already ranked first.

Two caveats: the gains are maximums measured on a research benchmark, not average results in production, and the engines have changed since. The direction holds: precise, quotable, well-sourced content is easier to reuse.

Being talked about matters more than being linked

Brands that are mentioned widely across the web are more visible in AI Overviews than brands that only have many backlinks.

Ahrefs studied tens of thousands of brands and found that branded web mentions had the strongest correlation with brand visibility in AI Overviews (0.664), well ahead of backlinks (0.218) ^[2]. Ahrefs itself points out that correlation is not causation. For a B2B brand, the practical reading is that third-party descriptions of what you do (reviews, comparisons, industry articles, partner pages) count, not only your own site.

The crawler settings that can remove you from answers

Each engine uses separate crawlers for training and for search. Blocking the wrong one takes you out of the answers.

User-triggered fetchers (ChatGPT-User, Perplexity-User) act on behalf of a person and, according to both vendors, may not follow robots.txt ^[3][6].

What this means for a B2B brand

• Check your robots.txt and CDN rules for the search crawlers before anything else. It is the cheapest fix there is.
• Write answer-first. Put the direct answer at the top of each section, with the figure and its source.
• State facts precisely and consistently (what you sell, for whom, integrations, compliance) on your site and on the third-party sources that describe you.
• Earn mentions, not only links: comparisons, reviews, partner listings and industry articles that name you.
• Measure before and after, with the same questions and enough runs to see past the noise.

Our AI technical audit covers the first point; the measurement protocol covers the last.