| Category | Function | Share | Rank | Negative rate | Labels | Quadrant | Since September 2026 |
|---|---|---|---|---|---|---|---|
| Business banking | Treasury and cash | 4% | 6 of 86 | 19% | 36 | accepted challenger | =heldSince September 2026: 2% → 2%, ±0 points. Inside the 10-point floor: within noise. Read over the models both editions asked. |
| Corporate cards | Spend and procurement | 0% | 57 of 67 | 50% | 4 | under 10 labels · led by Ramp at 52% | =heldSince September 2026: 0% → 0%, ±0 points. Inside the 10-point floor: within noise. Read over the models both editions asked. |
| Category | September 2026 | Now | Change | Reading | Rank |
|---|---|---|---|---|---|
| Business banking | 2% | 2% | =heldSince September 2026: 2% → 2%, ±0 points. Inside the 10-point floor: within noise. Read over the models both editions asked. | Within noise | Rank 9 → 6 of 86 |
| Corporate cards | 0% | 0% | =heldSince September 2026: 0% → 0%, ±0 points. Inside the 10-point floor: within noise. Read over the models both editions asked. | Within noise | Rank 59 → 57 of 67 |
Shares here are read over the models both editions asked, so they can differ by a point or two from the standing above, which counts every model in this edition.
The floor is 10 points of share, measured: how far the models move a leader on their own when the same questions are asked twice with nothing changed. A larger change is movement; a smaller one is noise, and both are shown. Movement is read over the twelve models both editions asked; GPT-6 Luna, Muse Glimmer 30B joined this edition and are in the standing but not yet in the comparison. How the floor is measured · The editions
| Model | First choice | Alternative | Mention | Negative | Labels |
|---|---|---|---|---|---|
| Claude Haiku 4.5 | 0 | 0 | 1 | 0 | 1 |
| GPT-5.4 mini | 0 | 3 | 0 | 1 | 4 |
| Gemini 3.5 Flash | 0 | 3 | 1 | 2 | 6 |
| Perplexity Sonar | 0 | 0 | 0 | 0 | 0 |
| Grok 4.1 Fast | 0 | 2 | 1 | 3 | 6 |
| Mistral Small | 0 | 0 | 3 | 0 | 3 |
| DeepSeek V4 Flash | 1 | 1 | 2 | 0 | 4 |
| Llama 4 Maverick | 0 | 0 | 1 | 0 | 1 |
| Qwen 3.7 Flash | 0 | 1 | 1 | 2 | 4 |
| Kimi K2 | 0 | 0 | 1 | 0 | 1 |
| GLM 4.7 FlashX | 0 | 1 | 2 | 0 | 3 |
| MiniMax M2.5 | 0 | 1 | 1 | 0 | 2 |
| GPT-6 Luna | 1 | 1 | 0 | 0 | 2 |
| Muse Glimmer 30B | 2 | 0 | 0 | 1 | 3 |
Verbatim evidence the judge attached to positive labels.
“Bank of America Global Commercial Banking with Business Advantage checking and CashPro treasury services for dedicated commercial relationship coverage.” Muse Glimmer 30B · Business banking · direct prompt · first choice
“Chase and Bank of America are the banks most cited for businesses that need branch access.” Muse Glimmer 30B · Business banking · comparative prompt · first choice
“Go with JPMorgan Chase or Bank of America as your primary bank.” DeepSeek V4 Flash · Business banking · direct prompt · first choice
“compare Chase and Bank of America first” GPT-6 Luna · Business banking · comparative prompt · first choice
Verbatim evidence attached to negative labels. A warning on a product with few labels is a warning; on a product with many, it is one voice among them.
“Widely criticized by small businesses for high monthly fees and rigid compliance requirements.” Gemini 3.5 Flash · Business banking · negative prompt · hard negative
“Long holds on business deposits (7+ days), payment issues, account freezes, poor support.” Grok 4.1 Fast · Business banking · negative prompt · hard negative
“BOA sees a high volume of CFPB complaints regarding account closures.” Qwen 3.7 Flash · Business banking · negative prompt · hard negative
“Avoid traditional banks like Chase/BofA if budget-tight” Grok 4.1 Fast · Business banking · budget prompt · hard negative
Citations exist only for the models that return a source list, five of the fourteen in this edition, so these counts come from 94 of the 109 answers that named Bank of America and are not a share of its labels.
271 of the 296 domain citations in answers naming Bank of America came from somebody else's page.
Pages are listed as the models cited them.
Search figures are US estimates from DataForSEO, read September 28, 2026; AI search demand is its modeled, directional estimate, not a count of queries to any assistant. The answers are this edition's. Two measurements side by side: neither is read as the cause of the other.
| Kind | Pages | Last 90 days | 2025-10 to 2026-09 | Latest | Categories named |
|---|---|---|---|---|---|
| Blog | 3 | 1 | 2026-07-31 | ||
| Report or ebook | 1 | undated | |||
| Comparison | 1 | undated | |||
| Template or tool | 1 | undated | |||
| Glossary or explainer | 1 | undated | |||
| News or press | 1 | undated |
Every page bankofamerica.com exposes, subdomains included. Kind is read from the address and title. The last 90 days, the latest date and the twelve months count pages by when they were published, from the site's feeds, a date in the address, or the page's own publication date, read from up to a hundred of its most recently changed pages; a page that says only when it last changed is counted in its kind but not in when, so the recent counts are a floor, and a kind none of whose pages gives a publication date reads undated. Read September 29, 2026.
An email the morning each edition publishes: where this product moved, where it held, and by how much against the noise floor. One address, confirmed by a click; a stop link in every email.
Already following? Everything you follow, with a stop for each.
What Bank of America's own pages state, read September 28, 2026: bankofamerica.com. A claimed page can correct any of them.
Claiming is free and changes nothing in the data. A claimed page shows a verified contact who is told when each edition publishes and when Bank of America's standing changes by more than the noise floor; the right to propose corrections to the vendor table, meaning names the judge wrote that should or should not read as Bank of America, applied by version and listed in the change log; and a one-line description supplied by the vendor and marked as such.
A new claim receives the current edition's vendor brief for Bank of America by email, built from the raw record of the edition. It shows: