Start With the Mutual-Need Statement
Every strong partnership begins as a sentence: they need our licence and 400,000 SME customers; we need their payroll-native underwriting and developer experience. Write yours before the first pitch meeting, and two things happen. The meeting stops being a demonstration and becomes a negotiation between parties who both know the trade. And the deals that should never happen reveal themselves immediately, because if the fintech does not need what you have, you are negotiating from hope, and hope is the most expensive negotiating position in banking.
Test the sentence for falsifiability: could the other side dispute it? They-need-our-customers, is that measured or assumed, and do they already serve your segments through other channels? We-need-their-technology, has anyone timed the build-versus-rent math, or is the technology attractive because building things is slow here? Learning how to assess a fintech partnership starts here because everything downstream, the checklist, the pricing, the verdicts, inherits the quality of this sentence.
Place the Fintech on the Value Chain Before You Talk Terms
Before terms are discussed, place the counterparty on the map. Which value chain link do they attack, Deposit, Lend, Pay, Invest, Insure, Advise, and what is the wedge, the specific customer moment they exploit? Then score them on the two axes that decide posture: competitive threat and partnership value. A player can be a genuine partner on aggregation while being a rising threat in payments; the grid exists precisely because both can be true at once, and the answer differs by link.
The scoring also disciplines the meeting itself. A fintech that presents as a partner while climbing toward your core, new licence applications, lending job postings, a roadmap edging from enablement to origination, is telling you its trajectory, and the grid is where you write that down. The working prompt for this scoring step is in our free digital banking prompts library; the assessment below assumes the posture question has been asked honestly.
The Seven-Item Checklist
Seven items decide most bank-fintech outcomes, and every one of them should be answerable from the term sheet. If an item cannot be answered from the paperwork, that is not a gap in the documentation, it is the deal's real shape, waiting to surprise you.
| Item | Question the paperwork must answer | Failure mode if vague |
|---|---|---|
| Customer ownership | Who books, contacts, and keeps the relationship at exit? | The customer walks with the partner |
| KYC and AML standards | Whose standard governs, who checks, who audits, who pays fines? | The fine arrives with your logo on it |
| Underwriting veto | Can the licence-holder enforce credit standards the platform cannot override? | A credit book you did not choose |
| Complaint and refund liability | Who routes it, within what timelines, who pays for platform-caused harm? | Reputation costs with no owner |
| Data ownership | Who keeps the transaction history at exit? | The partnership builds their asset, not yours |
| Kill criteria | Which named written conditions end the program? | A hostage situation with better branding |
| Regulator visibility | Does the structure survive the explain-this-program question? | A structure you cannot defend under examination |
Source: Compiled from the assessment checklist in this guide, ZLearning 2026
Control Items: Ownership, KYC, Underwriting
Item one, customer ownership: who books the customer in their systems, who may contact them, and who keeps the relationship if the program ends. Item two, KYC and AML: whose standard governs, who runs the checks, who audits them, and who eats the fine when they fail. Item three, the underwriting veto: if balance sheets are involved, can the licence-holder enforce credit standards the platform cannot override, because you carry the losses the platform's growth model creates.
These three are grouped because they share a failure mode: each is easy to concede in a demo-driven negotiation and expensive to win back after launch. Ownership conceded at signing is the customer relationship walking away in three years; a KYC standard borrowed from the partner is the fine arriving with your logo on it; a veto waived for speed is the credit book you did not choose.
Exit Items: Liability, Data, Kill Criteria, Regulator Visibility
Item four, complaint and refund liability: the routing, the timelines, and who pays for platform-caused harm. Item five, data ownership: who keeps the transaction history, and what happens at exit, this one decides whether the partnership builds your asset or theirs. Item six, the kill criteria: the named, written conditions that end the program, drafted while everyone is friendly, because a deal without written kill criteria is a hostage situation with better branding. Item seven, regulator visibility: draw the structure as the supervisor will see it, the licence as the meeting point, the accountability lines named, and check the structure survives the explain-this-program question.
Grouped together, the exit items answer one question: what does this look like when it ends badly? Partnerships end more often than their sponsors expect, strategy shifts, licenses change, growth outruns controls, and the deals that end well are the ones whose endings were designed at the start.
Pricing the Risk Into the Deal
The embedded-finance ledger most institutions eventually learn: the experience owner captures the customer moment and most of the margin; the enabler takes platform fees; the licensed bank earns interest and interchange minus program costs, minus the compliance burden and the risk reserve nobody modeled. The assessment's financial job is to make that reserve explicit before signature: the audit rights, the monitoring overhead, the remediation capacity, and the reputational cost of the newspaper test.
Two structural protections earn their negotiation time. The underwriting veto, you carry the balance sheet, so you own the standards, and staged economics, where revenue share grows with the partner's demonstrated control quality rather than with volume alone. Both convert the partnership from rent-our-licence to meet-our-standards-as-you-scale, which is the only version that survives its own success.
The Climb Test: Watching a Partner Become a Competitor
Partnerships age in one of two directions, and the grid's climb indicators tell you which. A partner moving up the value chain signals through observable facts: licence applications filed with regulators, job postings for lending and credit-risk roles, product pages that quietly add origination, earnings language that reframes your customers as its market. None of these is disloyalty, they are a business model behaving normally, but each one changes the exchange assessment, and the exchange is the whole deal.
Build the climb test into the quarterly review rather than waiting for the annual strategy offsite: list the four indicators, check them against public information in an hour, and score the drift. A partner holding its lane earns deeper integration; a partner climbing earns renegotiated terms, defensive product work, or a managed exit. The mistake is not partnering with potential competitors, it is being the last to notice what they became.
The Newspaper Test
The newspaper test prices the risk no line item captures. Write the front-page story of the partnership failing at its worst realistic failure mode, with your bank's logo attached, because that is how it prints: the supervisor's question, the customer harm, the program that outran its controls. Then ask whether the expected revenue could fund that headline: fines, remediation, trust repair, and the internal hours of the response nobody budgets.
The test has three outcomes. If the economics comfortably fund the headline, the risk is priced and the deal can proceed. If they roughly cover it, the risk pieces, veto, audit rights, kill criteria, need renegotiating until the margin returns. If the headline dwarfs the revenue, walk: the market will offer other partnerships, but the licence and the trust are non-renewable. Walking away from mispriced risk is a core competence, and committees should celebrate it as one.
The Market Numbers Behind Build, Buy, and Partner
The partnership decision sits inside large numbers on every side. Gartner estimated worldwide banking and investment-services IT spending at roughly $652 billion in 2022, the build-it-yourself baseline is enormous. On the other side, McKinsey projected embedded finance revenues reaching about $230 billion by 2025, and BCG's embedded finance research put the transaction value flowing through such models on a path toward $7.2 trillion by 2026. The numbers explain both the deal flow and the urgency: the margin pool is real, and so is the cost of building it alone.
Sources: Gartner IT spending forecast 2022 · McKinsey embedded finance projection · BCG embedded finance research
What the Capability Race Does to Your Negotiating Position
A quieter statistical shift changes the terms of trade inside every deal. A joint Bank of England and FCA survey found about 72% of UK financial firms using or developing AI, McKinsey places regular generative-AI use at roughly two-thirds of organizations, and Citigroup's 2024 analysis found about 54% of banking jobs have high automation potential. A bank whose staff can score a partner's model, interrogate its data practices, and build the alternative in-house negotiates from evidence; a bank dependent on the demo negotiates from admiration.
Sources: Bank of England and FCA AI survey 2024 · McKinsey State of AI 2024 · Citigroup GPS report 2024
How to Read the Two Charts Together
The dollar chart says the partnership lane is where the growth economics live; the capability chart says the price of entry is internal competence. Together they reframe the build-versus-partner question: the institutions negotiating the best embedded-finance terms are those whose people can evaluate the technology with their own tools and frameworks, the four-layer working stack in our guide to the best AI tools for banking professionals exists partly for this reason. Capability is not a substitute for partnering; it is what makes the partnership's terms survivable.
The Three Verdicts: Proceed, Renegotiate, Walk
The assessment ends in one of three answers. Proceed: the mutual need is real and falsifiable, the seven items are contractually solid, and the newspaper risk is priced with room to spare. Renegotiate: the value is real but the risk allocation is not, the veto, the audit rights, the kill criteria are the pieces to win before signature, and staged economics is the currency to trade. Walk: the headline risk exceeds the economics, or the mutual need turns out to be one-sided, or the climb trajectory says you are funding your next competitor.
All three answers are successes; the only failure is proceeding with the exposures unpriced because the demo was impressive. Write the verdict down with its reasoning, verdicts that live in meeting memories get revised quietly, and revised verdicts are how banks end up running programs everyone privately doubts.
| Verdict | Conditions that trigger it | First move |
|---|---|---|
| Proceed | Mutual need falsifiable, seven items contractually solid, headline risk priced with room | Sign with staged economics and audit rights in place |
| Renegotiate | Value is real but the risk allocation is not, veto, audit rights, or kill criteria weak | Trade staged economics for the risk pieces before signature |
| Walk | Headline risk exceeds economics, need is one-sided, or the partner is climbing | Exit politely and log the reasoning for the next deal |
Source: Compiled from the three-verdict method in this guide, ZLearning 2026
Quarterly Reassessment: The Discipline After Signature
After signature, the discipline shifts to rhythm. Quarterly: is the mutual need still mutual, are the checklist items still solid as the contract is operated rather than as it was signed, what do the climb indicators say, and are the kill criteria still written in a form everyone remembers? Annually: read the kill criteria aloud, re-run the newspaper test against current volumes, and re-score the grid, posture is a living variable, not a signature-day artifact.
Partnerships drift quietly because every individual change looks reasonable: a new product line, a new data use, a renamed team. The scheduled review is how you notice drift before the supervisor does, and the fact that you noticed first is itself worth a great deal in the conversation that follows.
The 30-Day Assessment Plan
The method installs in a month and leaves behind a reusable instrument. Run it on the deal nearest signature, then run it on every deal that follows:
- Days 1-5: write the mutual-need statement for each live counterparty, and mark which statements survive a falsifiability test
- Days 6-10: place each fintech on the value chain and the threat/partner grid, and record the climb indicators you will track
- Days 11-20: run the seven-item checklist against the nearest term sheet, and list every item that cannot be answered from paperwork
- Days 21-25: run the newspaper test on the worst realistic failure mode, and price it against expected program revenue
- Days 26-30: issue the verdict, proceed, renegotiate, or walk, in a one-page initiative brief the committee can approve in one sitting
The Standard Worth Holding
The method reduces to a habit: one sentence each side can dispute, seven items the paperwork must answer, one headline you are willing to fund, and a verdict written down by someone accountable. If you want the surrounding frameworks, the value chain, the grid, the audit that tells you what you actually need from a partner, Day 1 of our digital banking course is free, no card, and it produces the baseline the assessment runs on. Deals will keep arriving; the discipline is what decides which ones deserve you.
Frequently asked questions
▸What is the first question to ask about any fintech partnership?
Not what-does-the-product-do but what-does-each-side-need-from-the-other, the mutual-need statement, written in one sentence before the first meeting. A bank brings licence, balance sheet, customers, and trust; a fintech brings a moment, technology, audience, or model. If you cannot write the sentence, if only one side demonstrably needs the other, you are not looking at a partnership; you are looking at a vendor relationship or a competition, and the posture should change accordingly.
▸Which contract terms matter most in bank-fintech deals?
Seven, in rough order of how often they decide outcomes: customer ownership (who books, contacts, and keeps the relationship), KYC and AML standards (whose rule, who audits), underwriting veto (whether the licence-holder can enforce credit standards), complaint and refund liability routing, data ownership at exit, the kill criteria (named, written conditions that end the program), and regulator visibility, whether the structure survives a supervisor's explain-this-program question. Revenue splits get the negotiation attention; these seven get the regret.
▸What is the newspaper test for a fintech deal?
Write the front-page story of the partnership failing at its worst realistic failure mode, with your bank's logo attached, because that is how it prints. The supervisor's question, the customer harm, the program that outran its controls. Then ask whether the expected revenue could fund that headline: fines, remediation, trust repair. If not, the deal is mispriced, renegotiate the risk pieces (veto, audit rights, kill criteria) or walk. Walking away from mispriced risk is a core competence, not a missed opportunity.
▸How often should an existing partnership be reassessed?
Quarterly, with the same discipline as any strategy instrument: is the mutual need still mutual, are the checklist items still contractually solid, has the partner's threat profile changed, the grid's climb indicators such as new licence applications or lending job postings, and are the kill criteria still written in a form everyone remembers? Partnerships drift quietly; the quarterly review is how you notice before the supervisor does.
▸What is the difference between a fintech partnership and a vendor relationship?
Mutual need and mutual exposure. A vendor sells you software and carries none of your regulatory obligations; a partnership puts two balance sheets, two brands, and two sets of customers inside one customer journey, which is why the seven-item checklist exists. The practical test: if the arrangement can be cancelled with a notice period and no supervisor would ever ask about it, you have a vendor. If the fintech touches your customers, your licence, or your balance sheet, assess it as a partnership or do not do it.
▸What should a bank never outsource to a fintech partner?
Accountability, even where execution can move. The licence obligations, the KYC and AML standard, the credit policy the balance sheet answers for, and the complaint-handling duty to customers must remain owned by the regulated entity even when a partner performs the work. Execution can be delegated; responsibility cannot, supervisors consistently treat the licence-holder as answerable, which is why the checklist insists on audit rights and a veto rather than trusting the partner's platform to enforce your standards for you.
▸Where do I start if my bank has no partnership framework?
Start with the one-sentence mutual-need statement on your live pipeline, it takes an afternoon and immediately sorts real partnerships from vendor conversations. Then run the seven-item checklist on the deal closest to signature. If you want the surrounding structure, Day 1 of our digital banking course is free and covers the value chain and threat grid that frame these decisions, and the one-page initiative method carries the eventual recommendation into a committee document.