Capital One and Chase are the only banks AI names first more than four times.

Two banks hold the first answer when AI recommends a bank. Capital One comes first in 12 of 60 recommendation moments. Chase comes first in 11. No other audited bank comes first more than 4 times.

Many banks crowd the board below them. Scores run from 81 to 60, and the median is 69. Every bank from Capital One down to Chime rates strong. U.S. Bank scores 72 and Chime scores 71. AI names neither of them first in any moment.

The gap also runs the other way. AI under-rewards the brand equity of PNC, USAA, Navy Federal Credit Union and TD Bank. Navy Federal Credit Union scores 18 on Clarity and 4 on AI Recommendation.

The shelf reaches past the audited banks. AI made 350 choices across 100 brands. The 15 audited banks hold 55.7% of them. American Express, Axos Bank and Charles Schwab take answers from outside the audited set.

ABOUT THIS READING

ARA reads each bank on five attributes: Findability, Clarity, Sentiment, Identity Fidelity and AI Recommendation. Each attribute is scored out of 20, and the total is out of 100. The tiers are AWESOME 83-100, STRONG 70-82, AVERAGE 56-69, WEAK 40-55 and INVISIBLE 0-39. A SWELL position means AI under-rewards the bank's real equity. EBB means AI recommends the bank above its equity. ALIGNED means the two agree.

A setup defect made Claude answer 8 of 15 recommendation questions without its web tools. ARA collected Claude again on 2026-09-28, and every published Claude answer comes from the corrected call. Earlier readings used the old call. Claude searched on those earlier batteries, and ARA found no sign of the defect there. But the call itself differs from this quarter's call. Part of any movement since last quarter may come from that change.

Winners and losers

Capital One leads the board at 81, up 16 points since last quarter. Sentiment rose 6 points and Findability rose 2. Every bank in the cohort rose 5 to 9 points on Sentiment, so Capital One's Sentiment gain sits inside a cohort-wide shift. Part of the movement may also come from the corrected Claude call. AI chose Capital One in 36 of 60 moments, across all four AI models.

Capital One is strongest on first checking accounts, students and young adults, building credit, customer service and trust. AI does not name it for gig workers or mortgages.

PNC made the biggest rise, up 24 points to 69. Clarity went from 10 to 18, and Sentiment went from 8 to 15. Part of the movement may come from the corrected Claude call. AI chose PNC in 7 of 60 moments, across three of four AI models. It is strongest on students and young adults, and AI never names it first there. AI does not name it for first checking accounts, switching banks, retirees or high-yield savings.

PNC placed three bets last quarter. The 'Brilliantly' bet aimed to change how GPT frames PNC. All four AI models now quote the slogan. But no question measures GPT's framing, and the identity reading has no comparable prior, so ARA grades the bet as no clean read. The national identity bet aimed at the same framing. PNC opened its 50th new branch and converted FirstBank, but no model mentions either one, so the grade is no signal.

The institutional voice bet aimed at Sentiment. Sentiment went from 8 to 15. Every bank in the cohort rose 5 to 9 points on Sentiment, so ARA cannot credit any of that gain to PNC. The grade is no clean prior. ARA credits none of the three bets with the Clarity gain from 10 to 18.

TD Bank sits at the bottom at 60. This is its first reading in this cohort, so it has no movement figure. AI chose TD Bank in 4 of 60 moments, across three of four AI models. No question gets it named by two or more AI models. AI does not name it for first checking accounts, switching banks, building credit or gig workers. It holds a SWELL position, so its equity is ahead of what AI rewards.

Bank of America made the smallest gain of the 8 returning banks, tied with Wells Fargo. It went up 7 points to 74. Findability fell from 14 to 13. Part of the movement may come from the corrected Claude call. AI chose Bank of America in 30 of 60 moments, across all four AI models.

Bank of America is strongest on mortgages, trust, first checking accounts, switching banks, and students and young adults. AI names it first only for mortgages. AI does not name it for gig workers, high-yield savings or customer service.

Bank of America placed three bets last quarter. The Erica bet aimed at Clarity, and Clarity rose 3 points. AI still quotes 2025 Erica figures and does not mention the July generative upgrade. ARA grades this bet as partial.

The trust content bet aimed at reputation, and the reputation reading rose 5 points. ARA could not link the rise to the bet. Every bank in the cohort rose 5 to 9 points on Sentiment. Fees, frozen accounts and the Epstein settlement still lead the outside record. The grade is no clear link.

The agentic gateway bet aimed at Findability. Nothing shipped. The site still returns 403 to agents, and agent readiness is still zero. Retrievability fell 1 point, and Findability fell from 14 to 13 in the same quarter. The grade is still zero. Only the Erica bet shows a partial link, on Clarity.

The Leaderboard.

AI spreads its banking answers across 100 brands

AI made 350 choices across 100 brands. The 15 audited banks hold 55.7% of them. The largest single holding is 34 choices, or 9.7% of the shelf.

American Express has the most choices outside the audited set, with 10. Axos Bank and Charles Schwab hold 9 each, and Marcus by Goldman Sachs holds 7. Bluevine, Found, Regions Bank and Rocket Mortgage hold 4 each. These brands compete for the same answers as the audited banks.

The tail is long. AI names 56 brands exactly once, and they hold 16% of all choices. AI names 15 brands exactly twice, and they hold 9%. Together, 71 of the 100 brands hold 25% of the shelf.

15 buyer questions x 4 models / audited brands in colour, every other brand in a stepped grey. Toggle between Cohort Only and Full Shelf.

KEY FINDING 01

Bank of America is named in half of all moments and comes first once.

A maximum score on AI Recommendation does not mean that AI names the bank first. Capital One and Chase both score 20. Capital One comes first 12 times in 60 moments, and Chase comes first 11 times. Chase is named in more moments, with 37 against Capital One's 36.

Bank of America scores 19 and is named in 30 of 60 moments. It comes first once. Ally scores 19 and comes first 4 times. SoFi scores 18 and comes first twice. A bank on the list does not lead the list.

The leaderboard ranks banks by total score. First-place counts give a different order. Capital One leads both. Bank of America is fifth on score and is named in more moments than any bank below the top two.

KEY FINDING 02

Navy Federal Credit Union scores 18 on Clarity and AI names it in 5 of 60 moments.

Four banks hold a SWELL position: PNC, USAA, Navy Federal Credit Union and TD Bank. Each scores 16 or higher on Clarity. Each scores 8 or lower on AI Recommendation. TD Bank scores 17 on Clarity and 3 on AI Recommendation.

Four banks sit on the other side. Chase, SoFi, Ally and Bank of America hold an EBB position. AI recommends each of them above its brand equity. Bank of America scores 12 on Identity Fidelity and 19 on AI Recommendation.

The Brand Fundamentals vs Recommendation grid shows each bank's equity against its recommendation reading. The SWELL banks have the clearest opening. Their fundamentals already support more recommendation than AI gives them.

Brand Fundamentals vs Recommendation.

KEY FINDING 03

U.S. Bank rates strong at 72 and AI never names it first.

U.S. Bank has the highest Findability on the board, at 15. It scores 15 on Sentiment and 15 on Identity Fidelity. Its AI Recommendation score is 11. AI names U.S. Bank in 9 of 60 moments and never names it first.

Chime shows the same pattern. It scores 71, with 18 on Clarity and 15 on Identity Fidelity. AI names Chime in 12 moments and never names it first. Both banks hold an ALIGNED position, so ARA reads their recommendation as consistent with their equity.

KEY FINDING 04

Wells Fargo, Citi and Bank of America share the lowest Identity Fidelity on the board, at 12.

Every audited bank scores 15 or higher on Clarity. Capital One, Chime, PNC and Navy Federal Credit Union score 18. Identity Fidelity runs lower, from 12 to 15.

Three large national banks sit at the bottom of that range. Wells Fargo, Citi and Bank of America each score 12 on Identity Fidelity. AI describes these banks clearly. It keeps less of what makes each one distinct.

Ally, U.S. Bank, Chime, PNC, USAA and TD Bank score 15 on Identity Fidelity, the top of the range. The Understood vs Distinct chart places each bank on both readings.

Understood vs Distinct.

KEY FINDING 05

American Express wins 10 banking choices from outside the audited set.

The audited banks hold 55.7% of the 350 choices. Brands outside the set take the rest. American Express holds 10 choices, Axos Bank 9, Charles Schwab 9 and Marcus by Goldman Sachs 7.

The Cohort Only view shows the audited banks alone. The Full Shelf view adds the outsiders and the long tail. In the Full Shelf view, 71 of the 100 brands hold 25% of the shelf. Each of those brands appears only once or twice.

WHAT THIS MEANS

Capital One and Chase hold the first answer in banking AI today. Every other audited bank comes first 4 times or fewer. A strong total score does not move a bank to first place by itself. U.S. Bank and Chime both rate strong and never come first.

The SWELL banks have the clearest opening. AI does not yet reward the equity of PNC, USAA, Navy Federal Credit Union and TD Bank. PNC rose 24 points, but ARA cannot credit the rise to its bets. Every movement figure also includes the change in the Claude call. The next reading will show which gains hold.