Analysis of more than 1,200 scored AI answers shows the same brands, asked about with the same buying questions in the same month, were mentioned in 81% of one AI engine’s answers and just 43% of another’s — a gap invisible to companies that only spot-check ChatGPT.

Treyci, an AI visibility intelligence platform, released measurement data showing that major AI engines disagree sharply about which brands to mention and recommend when buyers ask for purchasing guidance. In one measured B2B software category, one AI engine referenced the tracked brands in 81 percent of its answers while another referenced the same brands in only 43 percent — a nearly two-fold visibility gap between engines that buyers use interchangeably.

The findings come from Treyci’s measurement methodology, which runs approximately 100 buying-intent questions per category — “best X for mid-size teams,” “X alternatives,” pricing comparisons — across ChatGPT, Perplexity, Gemini, and Grok, repeating every prompt three times per engine each month and scoring more than 1,200 resulting answers for brand mentions, recommendations, and citations.

Additional findings from Treyci’s recent measurement work include:

• AI answers are probabilistic. The same engine, asked the same buying question in separate sessions, routinely returns different vendor lists — meaning single-run brand checks capture an anecdote rather than a measurement.
• AI engines cite third parties, not vendors. When engines answer buying questions, cited sources are dominated by review platforms, comparison articles, and industry publications rather than company websites — reshaping where brand-visibility work actually pays off.
• Adoption is ahead of measurement.  In a Treyci scan of 100 B2B SaaS companies, 41 published an llms.txt file for AI crawlers, yet few companies can quantify whether any of their AI visibility work has changed how often engines recommend them.

“Marketing teams are making decisions about AI search based on one screenshot of one answer from one engine,” said Keith Schilling, founder of Treyci and previously an AEO/GEO practitioner at PayPal. “The data says that’s a coin flip wearing a suit. The engines disagree with each other, and they disagree with themselves from one asking to the next. Until a brand measures the distribution of answers — across engines, on repeat runs — it doesn’t know its own numbers.”

The shift matters because AI assistants increasingly produce the first shortlist in B2B purchasing decisions. When a brand is absent from an AI answer, the company’s analytics show nothing: no impression, no session, and no record that a buying conversation occurred.