66% of Vendor Lists in Fintech AI Answers Put Their Own Product First

Vendors wrote 7 in 10 of the comparison pages ChatGPT and Google AI Mode cited when shortlisting fintech providers, and two in three of the vendor lists we could read put the vendor first. SEOs have argued all year about whether Google now punishes this kind of self-promotion. In AI shortlists of fintech providers, such lists turned up in most answers from both engines. That changes which comparison pages a fintech should publish, and which rival pages it should watch.

The short version

• Vendors wrote the lists: 7 in 10 comparison pages that ChatGPT and Google AI Mode cited for fintech shortlists came from a vendor's own website (233 of 327).
• Most put themselves on top: Two in three vendor-written lists cited for fintech shortlists put the vendor's own product first (116 of the 177 we could read).
• Both engines cited them: A vendor list that put its author first appeared in at least 45 of 60 AI Mode and 34 of 60 ChatGPT answers shortlisting fintech providers.
• Business banking was the exception: There, publishers such as NerdWallet and Forbes had 37 of 69 comparison-page citations, and vendors had 21.
• AI Mode brought other vendors into brand questions: When a question named a fintech, AI Mode cited another vendor's comparison page in 36 of 60 answers, and ChatGPT in 2.

What to do with this

1. Where vendor lists dominate the category, publish the brand's own. Outside business banking, vendors had 4 in 5 comparison-page citations (281 of 350). Answers that cited a fintech's own list also recommended it far more often, though that link may run either way. Track the category's shortlist prompts before and after publishing. In business banking, start with point 3.
2. Where AI Mode matters in the brand's market, audit the rival pages that name the brand. Ask it how the brand compares with each rival, what it costs, and whether it is licensed. Log every other vendor's page it cites. Where a page gets fees, features, or licensing wrong, put the correct facts on the brand's pricing page and its own [brand] vs [rival] page.
3. In business banking, get onto publisher lists. Send the editors of the cited lists current fees, limits, and licensing details. This rests on 5 prompts, so test it on the brand's own tracked prompts before moving budget.

Vendors wrote 7 in 10 cited comparison pages, and 2 in 3 of those we could read put the vendor first

Of the 327 comparison pages cited in shortlist answers, vendors wrote 233. Most were not the fintechs we studied. Other software firms, payment companies, and even banks wrote 167 of them.

SAP Concur's "Best Expense Management Software for Businesses in 2026," cited 4 times, puts Concur first of six, and that was typical: of the 177 vendor lists we could read, 116 put the vendor on top.

Our read: We think these pages get cited because they answer the exact question a finance lead types. We only saw the lists the engines cited, not the ones they skipped, so we can't say whether either engine prefers self-promotion.

Own lists also went with recommendations. When an answer cited a fintech's own list, it recommended that fintech in 41 of 62 cases. Otherwise it was about 1 in 4 (270 of 1,138). That is a correlation, not proof that the list did it.

Two caveats. Of the 116, 18 call their list unranked or say no single product is best, and we counted them as first anyway. And if none of the 56 vendor pages we couldn't read put the vendor first, the share would fall to 1 in 2.

In business banking, publishers had 54% of comparison-page citations

NerdWallet's "Best Banks for Startups 2026" and Forbes's "Best Business Checking Accounts Of 2026" were each cited twice. Across business banking shortlists, publishers and comparison websites had 37 of the 69 comparison-page citations.

Our read: We think banking differs because consumer finance publishers already review business accounts, so the engines have independent lists to cite. The 2 US banking prompts split the same way, with vendors at 6 of 23 citations. Five prompts is thin, so treat this as a pattern to test.

AI Mode cited another vendor's comparison page in 36 of 60 answers about named fintechs

Some questions named one or two fintechs and asked how they compare, what they cost, whether they're licensed, or what reviewers say. When asked BILL vs Melio for a small business paying vendors, AI Mode cited Tipalti's and Wise's own Melio vs BILL pages on both days. ChatGPT cited neither.

Our read: We think AI Mode answers a two-company question much as it answers a shortlist question, with pages from across the category. On X vs Y questions, any vendor that has written its own page on the pair can win the citation. These 60 answers come from 30 prompts asked on two days, so each question counts twice.

Methodology

Question. Who publishes the comparison pages ChatGPT and Google AI Mode cite for fintech shortlist and vendor-check questions, and where do vendor-written lists place their own product?

Data. The Omnius fintech visibility study covers 60 B2B fintechs in six segments and 90 prompts. Each ran on ChatGPT and Google AI Mode through Atomic AGI's tracking (US run location, single turn) on October 27 and 28, 2026. Of the 30 shortlist prompts, 11 describe US companies and 6 UK ones; the rest cover Germany, France, the Netherlands, and five other markets. The 5 business banking prompts cover the US (2), the UK, France, and the Netherlands. This note uses its 60 shortlist and vendor-check prompts (240 answers, 1,967 citations).

Sample. 419 comparison-page citations in shortlists and 215 in vendor checks, counted each time an answer cited a page. Video, forum, review-platform, app-store, Wikipedia, regulator, news, and social pages are excluded.

Coding. Comparison pages are best-of lists, X vs Y pages, and alternatives pages, identified by list words in the title or address. Websites were coded from their homepages as vendor, service firm, publisher or comparison website, or unclear (3 of 634 citations). Of the 233 shortlist vendor pages, 34 were cited only by homepage address. The other 199 were opened on November 9, 2026, and 177 loaded and listed products; each was coded with AI-assisted reading against a written protocol. A second, independent AI-assisted pass recoded 25 random pages (seed 1109), without the first codes, and agreed on 24.

Analysis. Intervals are 95% bootstrap intervals over the 30 prompts per stage. Vendors had 72.1% of shortlist comparison-page citations (63.8% to 78.9%); self-first vendor lists had 37.7% (158 of 419; 30.9% to 44.7%). Recommendation rates pair each shortlist answer with the 10 fintechs in its segment (1,200 pairs); the gap between 66% and 24% is 27.8 to 58.2 points. Citations and pairs cluster within prompts, so page-level shares carry no interval.

Limitations. Two days, two engines, one US run location, and model versions not logged. Pages were read twelve days after the runs.

Reproducibility. Aggregates are in each chart's data table. Prompt wording and answer text stay internal, apart from the BILL vs Melio example.

Sources

1. PPC Land, "Vendor listicles gain 38% median traffic since Google's January demotion," August 25, 2026, reporting Kevin Indig's Growth Memo study of August 24, 2026. https://ppc.land/vendor-listicles-gain-38-median-traffic-since-googles-january-demotion/

Omnius articles this note builds on

• Fintech AI Visibility Study: What ChatGPT and AI Mode Cite by Stage