New Angelfish Marketing Study of 1,052,053 AI Citations Reveals What AI Reads to Recommend B2B Software

Analysis of over a million AI citations maps the sources AI assembles its answers from for B2B SaaS brands

Angelfish Marketing, a UK B2B digital marketing agency specialising in answer engine optimisation (AEO) and B2B SaaS SEO, today published original research analysing 1,052,053 AI citations to map what AI assistants actually read when they recommend B2B software. Rather than asking whether brands appear in AI answers, the study takes the answers apart to show what they are built from.

The finding: AI assembles its answers about B2B software overwhelmingly from third-party sources, not the brands’ own websites. Across the answers that mention a brand, the brand’s own site is cited as a source only around a quarter of the time (13–32%). The rest is built from competitors, community platforms, review sites and the trade press — a landscape most brands neither own nor actively influence.

“Marketers assume that if AI is talking about them, it’s reading their website to do it,” said Dom Moriarty, Head of Growth at Angelfish Marketing. “It mostly isn’t. It’s reading what everyone else has said about you — the reviews, the Reddit threads, the YouTube videos, the roundups. Your own content earns you the right to be in the conversation, but it’s these third-party sources that actually make up the answer.”

Marketing Technology News: MarTech Interview with Miguel Lopes, CPO @ TrafficGuard

The study found the sources AI relies on form a knowable, finite set for B2B SaaS brands. Community platforms are the largest reachable group, with YouTube and Reddit the two most-cited individual sources across the entire study by a wide margin, and LinkedIn third — ahead of every review site and publication. Medium, review platforms such as G2 and Capterra, and trade titles such as TechRadar and Forbes recur across every brand studied. In total, 42 third-party sources appeared consistently across all six brands: the sources that shape how AI describes B2B software.

Part of the reason so much comes from third parties lies in how AI answers B2B software questions: by comparison. When a buyer asks something like “the best CRM for a small sales team,” AI weighs several products against each other rather than describing one in isolation, because that is how B2B buyers evaluate software. Brands appear alongside their competitors by default, drawing on the sources that discuss the whole category.

Marketing Technology News: Disrupt or Be Disrupted: The AI Wake-Up Call for B2B Marketers

This points to what Angelfish describes as the two jobs of AI visibility. Owned content and technical SEO earn a brand eligibility to be cited at all; presence across the third-party sources AI actually reads is what earns a place in the answer. Most growing brands, the study notes, have built only the first.
“Publishing more of your own content is necessary, but it’s the start of the job, not the whole of it,” Moriarty added. “The brands winning AI visibility have built a presence across the sources AI trusts — the reviews, the communities, the earned media. That’s the work most companies haven’t started, and it’s exactly where answer engine optimisation comes in.”

The study is descriptive rather than causal, and is based on data from the SEMrush AI Toolkit, blended across ChatGPT, Google AI Overviews, Google AI Mode and Gemini. It covers six established B2B SaaS platforms across CRM, project management and marketing automation, two brands in each category.