If you are evaluating a Banking-as-a-Service provider right now, you are probably six weeks into a process that will take six months and end with the wrong answer.
I have watched this happen often enough to be able to predict the failure mode. Founders run a long, polite, demo-driven evaluation. The vendors do what vendors do: they reframe questions, shift the conversation to roadmap features, and let the founder confuse motion with progress. The founder eventually signs because the calendar has moved and a launch date is committed. Eighteen months later, the founder discovers that they do not own the sponsor bank relationship, the interchange economics are 40% worse than they thought, and the cost to exit is structurally prohibitive.
This guide is the structured 30-day evaluation I would run if I were standing up a new fintech tomorrow. It is opinionated, it ignores roadmaps, and it interrogates the four things that actually drive whether a BaaS provider is the right partner: who owns the bank relationship, where interchange flows, what reconciliation actually looks like in production, and what it costs to leave. Everything else is decoration.
TL;DR
- The BaaS sales cycle is engineered to take six months. Run a 30-day structured evaluation instead.
- Lock your scoring grid before any demo. Demos are designed to reframe your questions.
- The single highest-signal question: "Name the sponsor bank, and let me read the tri-party agreement."
- Get interchange splits in writing to the basis point, or walk.
- Stress-test reconciliation with one anonymized week of files before signing.
- Model the exit before you model the launch.
Why BaaS evaluations go wrong
The structural problem is that the vendor is selling you opacity and you are buying transparency. The BaaS provider's margin depends on you not understanding exactly where each basis point of interchange lands. That tension does not get resolved by being polite. It gets resolved by demanding specific paper before you sign, and walking when the paper does not appear.
A useful framing: a BaaS provider is a bundling intermediary between you and a chartered bank. Anything the provider does that the bank could not also do directly is service that you are paying for. Anything the provider does that the bank cannot do is the part of the bundle that creates value. Most founders never separate the two, and the price they pay reflects that.
The four economic primitives
Before any demo, you need to be able to name four things on a single sheet of paper:
- What you move. Cards (debit, credit, prepaid)? ACH? Wires? Real-time payments? Each is a different regulatory and economic stack.
- The unit economics you need to clear. What basis points of interchange must you retain, net of network fees, processor fees, BaaS fees, and the sponsor bank's take, for your business to work?
- The regulatory entities involved. Single sponsor bank? Two-bank model (issuer + sub-issuer)? Money transmission licenses required at the state level? An MTL strategy is a 12-month project; you cannot bolt it on later.
- What exit looks like. In 36 months, if you have grown to the point where direct bank and processor relationships are economically rational, what does the migration cost? Both in dollars and in calendar time.
If you cannot write these four things down before the first demo, you are not evaluating vendors — you are letting vendors evaluate you.
The 12-row scoring grid
The grid is the most important artifact of the evaluation. Build it before any sales call. Lock the rows. The vendors will try, on every call, to introduce new dimensions that they happen to be strong on. That is the trap.
A grid I have used for card programs:
| Dimension | Weight | Vendor A | Vendor B |
|---|---|---|---|
| Sponsor bank identity (disclosed up front) | 10 | | |
| BIN ownership and portability | 9 | | |
| Interchange split (in writing, by MCC) | 10 | | |
| Settlement timing (T+0 / T+1 / T+2) | 7 | | |
| Reconciliation files (samples available?) | 9 | | |
| Compliance posture (deposit account in bank's name?) | 10 | | |
| Tri-party agreement readable pre-signing | 8 | | |
| Exit cost (documented, not "we'll work with you") | 9 | | |
| Card production timing | 5 | | |
| Customer support model (who answers a fraud call at 2am?) | 6 | | |
| Reference customers I can call directly | 7 | | |
| Time to first transaction | 5 | | |
Notice: roadmap features are not on the list. Neither is dashboard quality. Those are decorations. Score on the structural decisions that you cannot reverse in year two.
The one question that does most of the work
If you ask one question on every call, ask this:
"Which sponsor bank is on the back of every card my customer holds, and can I read the tri-party agreement before signing?"
The answer separates BaaS providers into two categories. The first category names the bank in the first thirty seconds and offers the tri-party agreement under NDA. The second category routes around the question. They will say things like "we manage that relationship for you" or "we have multiple bank partners." That second answer is the brokering of opacity, and it is the single largest predictor that you will be unhappy at month 18.
There is a specific reason this question matters. In a BaaS relationship, the consumer's deposit account is held at the sponsor bank, and the sponsor bank is the entity with the regulatory relationship to your customer. If you do not know which bank that is, and if you have not read the agreement that governs your access to that bank, you do not control your business. You rent it.
Interchange transparency — in basis points, in writing
Every BaaS provider knows their interchange economics to the basis point. Many of them will refuse to write it down. That refusal is the answer.
A reasonable request before signing:
- A sample interchange schedule for your expected MCC mix
- The bank's take (typically 20–35 bps for a sponsor bank in this role)
- The BaaS provider's take (typically 30–60 bps depending on services rendered)
- Network fees (Visa or Mastercard assessment + transaction fees)
- Processor fees
- Your residual
If the vendor will not put this in a document your CFO can defend, sign with someone else. The economics will not improve after the contract is signed. They will only become more visible.
Reconciliation is the truth-teller
You can learn more about a BaaS platform from one week of its reconciliation output than from twelve hours of demos. Ask for, under NDA:
- Seven days of redacted daily ACH return files
- Seven days of redacted daily card settlement files
- Seven days of bank ledger position files
Then sit down with a finance person and try to reconcile them to within one cent. If you cannot — if columns are missing, formats are inconsistent, or amounts do not tie back — that is your year-two reality. A platform whose reconciliation is broken at the sample stage is a platform whose reconciliation will be broken at scale, and reconciliation breakage at scale is what gets fintechs taken offline.
The exit clause is the most-skipped step
The single most-skipped step in BaaS evaluation is modeling the exit. Founders sign because they want to launch. Three years later, when they have grown to the point where direct sponsor bank, BIN ownership, and processor relationships would compress their cost stack by 30–60%, they discover that the contractual exit cost makes the migration uneconomic.
Before signing, get a written answer to:
- What is the documented exit fee schedule?
- Does the BIN range port to a new sponsor bank, or does it terminate?
- What is the BaaS provider's role in customer notification?
- What does the contractual transition assistance period look like?
- What customer data is yours, and what is the provider's?
A vendor who has done this honestly with other customers has a written playbook. A vendor who has not will say "we'll work with you on that." The first is acceptable. The second is not.
How this maps to the rest of your stack
This evaluation framework is part of Chisel's broader argument that software is a relationship, not a product. The relationships your BaaS provider intermediates — to the sponsor bank, to the processor, to the card networks, to your end customer — are the actual asset your fintech is building. Every basis point of opacity in those relationships is a basis point of value transfer away from you.
If you want to think about this more structurally, see our pillar on composable fintech infrastructure and the related glossary entries on sponsor banking and interchange.
A 30-day evaluation that produces a one-page scorecard, a written interchange schedule, a redacted reconciliation extract, a tri-party agreement copy, and an explicit exit cost is a better basis for a five-year decision than six months of polite demos. Run the structured process. Sign the paper or walk.
— Darin Petty