We scored 8 pages on banking as a service pricing . One domain of seven published a number.

Why do BaaS providers not publish their pricing?

BaaS pricing is negotiated per programme and usually sits under an NDA, so it gets resolved in a sales call rather than on a page. The clearest evidence is the highest-ranking provider comparison in our set. It lists a pricing cell for seven platforms, and not one of the seven carries a rate.

Those seven cells, reproduced exactly as published in Backbase's "Banking as a Service Platforms: Top 7 Compared for 2026" (retrieved 2 August 2026):

1. 1.Backbase: "Contact Backbase for enterprise pricing."
2. 2.Stripe Treasury: "Transaction-based. Contact for volume pricing."
3. 3.Treasury Prime: "Revenue share with partner banks."
4. 4.Unit: "Per-account and transaction fees."
5. 5.Solaris: "Platform fees plus transaction-based pricing."
6. 6.ClearBank: "Transaction-based with volume discounts."
7. 7.Synctera: "Revenue share with partner banks."

Three of those cells name a pricing model. Four are an instruction to get in touch. Seven of seven give you nothing to put in a spreadsheet.

The page ranking first for the query does something subtler. Unit's "The ultimate guide to Banking as a service" has a section headed Economics, and it asks the reader five questions:

• "Is there a platform fee?"
• "Will you be charged on a per-account basis?"
• "What does it cost to process different kinds of payments?"
• "What fees will you be able to generate, if any?"
• "How will that add to your customer acquisition cost (CAC)?"

It answers none of them. Those are the right five questions, which is the point. A page that poses exactly what a buyer needs to know, then declines to answer, is the cleanest illustration of the gap we found.

Australia: the same audit, run again

Of the five Australian banking as a service providers Google's AI Mode names when asked who supplies AUD accounts and card issuing, none publishes a rate. Boldrails retrieved all six cited domains on 17 August 2026. The only one that publishes a price quotes it in euros and carries no Australian content at all.

On 17 August 2026 we ran the August audit again against Australia, and we widened it. Two sample frames, scored separately and reported separately. Frame A is the banking-relevant organic first page for the same two queries, the spelled-out one and the acronym one, captured in Australia in English. Frame B is new: the six domains Google's AI Mode cited when asked which providers supply banking as a service in Australia for fintech AUD accounts and card issuing. Frame B exists because the Australian buyer is not typing a keyword. Total on-ICP Australian search demand for this product is roughly ten searches a month, so the answer set is where the question actually gets asked.

Frame A. The Australian organic first page

Frame B. The domains the AI answer cites

What the two disclosers actually published

Corporate Alliance, the single discloser in Frame A, publishes its figures as worked examples rather than as anyone's rate: a platform fee of 10,000 dollars a month and a second example at 5,000, a per-account cost of around 10 cents per customer per month, a per-transaction range running from a few cents for local transfers up to 30 dollars for an international wire, card interchange of around 2.5 percent, and a revenue split example of 1.5 percent against 1.0 percent. Every one of them is unattributed. Its companion article does carry hard monthly numbers, ranging from zero to 29 dollars, but those are what an end customer pays for a business account at named Australian banks and fintechs, not what a platform pays a provider. Under our scoring rules that is the wrong buyer, the same classification the August wave gave to build-side and bank-to-customer figures.

Boxopay, the single discloser in Frame B, is the more useful of the two because its numbers are its own: 0.018 EUR per payment at its highest volume tier, a 1,000 EUR monthly infrastructure fee for hosting, and setup starting from 0 EUR, with deployment and extensions priced on request. It is the most transparent source we have scored in either wave. It also publishes no Australian content, which is why an Australian buyer following this citation is reading a European rate card.

The acronym collides with a different industry in each market

In the United States the acronym query returned meeting bots, backend hosting and backup software. In Australia on 17 August 2026 it returned electric vehicles: eight of the nine first-page results were Battery as a Service, battery leasing plans for cars, most of them Indian motoring coverage. One banking result made the Australian first page for the acronym. The homonym is real in both markets and it is a different homonym in each, so an acronym keyword strategy does not fail the same way twice. It has to be measured per market.

Two waves, two markets, one result: the disclosure gap is not an American artefact. In both samples the only sources publishing buy-side numbers were vendors with a commercial interest, publishing unattributed figures, and in both samples reserve was disclosed by nobody. Australia adds a second failure mode on top. The buyer there is not searching, they are asking, and the answer set they get is thinner than the search results.

The stack

Which of the seven fee lines get hidden?

A BaaS quote is assembled from seven lines, and providers disclose them in inverse order of how much they cost you. Setup and platform fees are the ones occasionally published. Reserve, per-account, per-transaction and revenue split scale with your business, and no audited domain puts a number on any of them.

Setup / implementation

One-time, paid before you process anything.

Platform / monthly

The recurring access fee. Often carries a minimum.

Per-account

Charged per end-user account you open. Scales with growth.

Per-transaction

Charged per payment. Varies sharply by rail and by direction.

Reserve

Funds held against risk. Not a fee, but capital you cannot use.

Minimum commitment

The floor you pay regardless of volume.

Revenue split

The provider's share of interchange or fee income you generate.

The asymmetry is the point. The two lines that are fixed and predictable get published. The four that compound as you scale do not.

There is a tell worth knowing about. Unit's guide, which ranks first, publishes plenty of numbers, but every one is about revenue rather than cost: Shopify earning "more than 73% of their revenue from merchant solutions", Toast's lending business generating "$14M of revenue per year", interchange capture of 1.5% to 3%. It is a well-built page. It answers "how much will I make" while the reader is asking "how much will I pay".

Why do the published figures disagree?

The one domain that publishes figures disagrees with itself, and understanding why matters more than the numbers do. Gemba is the only source in the set publishing buy-side rates. Three things about it are load-bearing.

1. 1.It is a vendor, not an observer

   Gemba sells BaaS. Its pricing article quotes "traditional" integration at heavy setup cost and long timelines, then presents its own platform as the thing that removes them. Its UK article publishes its own rate card in detail, including a £15 account review fee, zero monthly maintenance for UK companies, and partners retaining up to 70% of custom fee revenue. That is disclosure, but it is disclosure as marketing, and the competitor figures work as an anchor that makes the vendor's own price look small.

2. 2.The bylines do not hold together

   The domain's pricing article carries the author "AutoSEO" in its published metadata and shows no visible byline at all. Its ROI article carries that same "AutoSEO" metadata while the visible page credits "Alexander Legoshin" more than a dozen times, so the machine-readable author and the human-readable author contradict each other on one page. A third article is bylined "Gemba Team". Anyone can check this by viewing the page source and comparing the article:author meta tag against the byline printed on the page. One of the names attached to the only numeric source in this lane is an SEO automation tool.

3. 3.The figures contradict each other

   Within its single pricing article the domain gives setup cost as $50,000 to $200,000, as £25,000 to £100,000, and as £30,000 to £75,000. Those are three mutually inconsistent ranges for one line item, on one page. The same article asserts a UK banking licence capital requirement without a source. We are not repeating that figure, because a single unsourced vendor page is not a basis for a regulatory claim.

There is one more place buy-side numbers surface. A Q&A on sacra.com carries specific figures, a deposit account at "$10,000 a month" and a card programme at "an additional $5,000 a month". They come from an "Anonymous BaaS business development executive", the speaker prefaces them by saying "I wouldn't share any particular numbers", and full access to the site costs $50 a month. So the most concrete practitioner figures on the open web are anonymous, hedged by the speaker, and behind a paywall.

How does Boldrails quote banking as a service?

Boldrails quotes against the seven lines this page audits. Ask for any of them in writing before you sign and you will get it, including reserve and the pass-through schedule. We do not publish a rate card either, because pricing moves with volume, market and risk. Since publishing this audit we have also named our own lines, and doing so exposed a gap in the seven-line framework itself. That is set out below.

That is a commitment about disclosure, not a discount. The audit above measures one thing: whether a provider will tell you what the lines are. That is the part a buyer can actually hold a provider to, so it is the part we are willing to put in writing.

We left ourselves out of the census on purpose. The sample frame is the banking-relevant organic first page for the two queries, Boldrails does not rank there, and adding a row for ourselves would have broken the method. The table below is a commitment, not an audit result, and it is labelled that way.

Two more things decide the deal and are not fees: the exit terms with data-portability obligations, and which institution holds the funds. Both are answered in the same document. Boldrails holds the necessary licences required in the markets we serve, and onboarding takes 3 to 14 days, depending on your case.

Last updated: 17 August 2026. Written by Claude Igrow.