Does domain authority predict AI recommendations for embedded finance companies?

Below a Domain Rating of roughly 80, domain authority had no predictive power over AI recommendations in this study. The Spearman correlation was -0.03 across 13 companies under that threshold. Treasury Prime at DR 49 was recommended 19 times. Rapyd at DR 74 was recommended once.

Which embedded finance companies appear most in AI-generated answers?

In this July 2026 study, Stripe (DR 95) was recommended in 37 of 40 answers and named first in 23. Adyen (DR 88) was recommended in 31. Among companies below DR 80, Unit (20), Marqeta (19), and Treasury Prime (19) were recommended most frequently.

How did you measure AI citations for this study?

Ten buyer-phrased prompts were run once each in ChatGPT, Gemini, Claude, and Perplexity using fresh chats, default settings, US location, and English. All 40 runs were completed within a single collection window on July 14, 2026. A company counted as recommended when named as an option, not when mentioned in passing.

Do AI engines show their sources for vendor recommendations?

Two of the four engines showed no sources. Gemini and Claude named vendors in every answer without displaying a single citation. Perplexity showed sources in all 10 of its answers. ChatGPT showed sources in 7 of 10 answers.

How Embedded Finance Companies Appear in AI-Generated Answers

In a study of 40 AI-generated answers to embedded finance buyer questions, domain authority stopped predicting AI recommendations below a Domain Rating of roughly 80. Across the 13 companies under that threshold, the correlation between authority and AI recommendations was effectively zero (Spearman -0.03).

Authority stops predicting visibility below DR 80

Stripe (DR 95) and Adyen (DR 88) dominated AI recommendations. Remove them and the Pearson correlation collapses to 0.03, Spearman to -0.03. Treasury Prime at DR 49 was recommended 19 times. Rapyd at DR 74 was recommended once.

The leaderboard changes with the question

Stripe was named first in 15 of 16 generic category answers, but only 1 of 12 specialist answers. Unit was named first in 6 of 8 banking answers. Challengers won visibility only on the specific questions that matched their specialty.

What the engines cite as evidence

Gemini and Claude showed no sources across 20 answers. The sources Perplexity displayed were dominated by SEO aggregator blogs and vendor content, not press or research. Finix's content influenced how one engine framed a buying decision without Finix being recommended.

The answers contain errors buyers cannot see

Three of four engines produced identifiable entity errors. Gemini referred to Standard Connect instead of Stripe Connect. Perplexity conflated Galileo and Finix. Claude cited the retired Railsbank brand and stated precise business metrics without sources. ChatGPT had no observed entity errors.

What this means for fintech marketing leaders

Domain Rating does not predict AI recommendations below the giant tier. The winnable ground is the specific buyer question, not the category. The evidence layer behind AI answers is buildable structured comparison content.