Fintech Unit Economics: How to Evaluate a Fintech Business

To evaluate a fintech's unit economics, measure what each customer or each dollar of volume contributes after the costs that come with it, then compare that with what it cost to acquire. Five checks do most of the work: convert gross revenue to net; use contribution, not revenue, for payback and lifetime value; test churn and acquisition cost; ask who funds the volume and at what price; and check how much revenue depends on one partner, merchant or rule. A fintech that passes all five can carry its fixed costs; one that fails any of them is usually growing on borrowed time.

Affirm is the cautionary example. It listed on Nasdaq in January 2021 at $49 a share and reached a record close of $168.52 that November. Then rates rose, its cost of funds increased, loan buyers paid less and its shares fell about 90% from their peak, as reported. Three years later, in the fiscal year to June 2026, it reported about $4.3 billion of revenue and about $417 million of GAAP operating income. Its loans had not become dramatically safer. Its unit economics had changed. This guide, drawn from The Money Equation, is educational, not investment advice.

Step 1: Convert gross revenue to net

Payment companies report revenue in two very different ways, and confusing them is the most common error in fintech comparisons. Gross take rate is gross revenue divided by volume, and it includes interchange and network fees that a processor collects and passes straight to card issuers and networks. Net take rate subtracts those fees and shows what the company keeps.

On the book's stylized $100 credit-card purchase, the merchant pays $2.75, but the acquirer and processor keep only about $0.40. The rest goes to the issuing bank and the network.

PayPal's figure is not comparable with the others until the pass-through is removed. Stripe's list price for small US merchants, 2.9% plus 30 cents online, looks far higher than Adyen's, but most of it is interchange. Our case study on how Visa makes money shows where each slice of a card fee goes.

Then ask whether the take rate rose because value was added or because fees were. Fiserv cut its growth outlook in October 2025, partly over pricing at its Clover merchant business that management called unsustainable, and its shares fell about 44% in a day, as reported. A fee that was added rather than earned tends to reverse.

Step 2: Use contribution, not revenue, for payback

A customer's value is what they contribute after variable costs: processing, fraud, support, funding and credit losses. Revenue payback flatters every fintech.

The book's stylized neobank in Chapter 15 makes the point with invented numbers. It spends $150 to acquire a member who generates $240 a year of revenue at a 40% contribution margin, or $96 a year.

A company that reports revenue payback can look far more efficient than it is, and a few points of extra churn can wipe out most of the margin. Ask for payback by cohort on contribution, and check whether the cohorts' contribution covers fixed costs too: technology, compliance, audit, licenses.

Step 3: Test churn, primary accounts and acquisition cost

For a consumer fintech, three inputs decide almost everything. A stylized cohort in the book's research follows 10,000 accounts at a made-up interchange-funded neobank for three years.

Primary accounts matter most. Revenue follows the paycheck: a customer whose salary lands in the account spends on its card and uses its other products, while millions of dormant accounts earn almost nothing. That is why Chime built its whole product sequence, early pay, the SpotMe overdraft cushion and wage advances, around direct deposit. It reports average revenue per active member, about $257 in 2025, rather than per account opened.

Acquisition cost decides whether growth creates value. Paid acquisition booms and referral bonuses can push it up fast. Distribution is the subject of the book's Chapter 16, which argues that the channel also decides which customers arrive, and therefore which risks.

Step 4: Ask who funds the volume

For a lender, the unit is a loan, and its margin is:

Lender's margin = yield − cost of funds − credit losses − servicing − acquisition cost

Affirm manages to revenue less transaction costs (RLTC): revenue minus loss provisions, funding costs, processing and losses on loan sales, which it has guided at roughly 3–4% of volume. In 2022 RLTC fell before losses moved at all, because funding got dearer and loan buyers paid less. A lender that sells its loans is only as large as its buyers' appetite at today's rates.

Affirm's answer was to diversify funding: partner-bank origination, warehouse lines, securitizations and forward-flow sales, including Sixth Street's December 2024 commitment to buy up to $4 billion of loans over three years. Lenders that owned a bank charter, such as SoFi and LendingClub, could keep lending on deposits in 2022, while Upstart had to hold loans on its own balance sheet when buyers retreated. The book's Chapter 17 treats funding, along with credit and fraud, as the limit on how fast a lender can grow.

Ask also how losses are priced. Affirm charges no late fees, so it cannot offset losses with penalty revenue and must price them up front. A lender whose loss data cover only good years has not yet been tested.

Step 5: Check concentration and rule dependence

Affirm's 2020 prospectus showed that one merchant, Peloton, accounted for about 28% of its revenue in the fiscal year to June 2020, as reported. When Peloton's sales collapsed, so did that revenue. Affirm later spread itself across platforms such as Shopify's Shop Pay Installments and Amazon, though it lost Walmart to Klarna in 2025. Look for any single merchant, partner bank, platform or distributor large enough to matter, and ask what happens at renewal.

Then look for rules. Most US neobanks earn debit interchange through partner banks with less than $10 billion of assets, which are exempt from the Durbin cap and can earn roughly two to three times the regulated rate (practitioner estimates). In the stylized cohort above, moving to a regulated rate turns a modestly profitable cohort deeply negative. The Durbin rule was vacated by a North Dakota court in August 2025, and the Eighth Circuit affirmed on 26 October 2026. Revenue that depends on an exemption is, in the book's phrase, revenue on loan.

The partner bank is a dependency too. Its consent orders, capital and appetite are the fintech's own risks, as the collapse of the middleware firm Synapse showed in 2024. Our case study on what happened to Synapse explains why a fintech must reconcile its ledger with the bank's every day.

Same revenue, different customer: Brex and Ramp

Two companies earning the same fee from the same rails show how much the customer choice matters. Brex and Ramp both sold corporate cards on partner banks and earned commercial card interchange, which the Durbin amendment does not cap. Brex pitched access and perks to venture-backed founders and was valued at about $12.3 billion in January 2022. It agreed in January 2026 to be sold to Capital One for about $5.15 billion, less than half its peak private value. Ramp pitched savings and spending control to finance teams. When the 2022–23 downturn made every company want to cut costs, that pitch kept working. Ramp's valuation rose to $32 billion in November 2025 and $44 billion in June 2026, with more than $2 billion of annualized revenue, as reported.

The fee was identical. What differed was which customer each chose and whether the product kept earning when that customer's budget tightened.

A checklist you can reuse

The takeaway

Fintech unit economics come down to one inequality: a customer, or a dollar of volume, must contribute more after losses and funding than it cost to acquire, with enough left to pay fixed costs. Convert gross to net, measure contribution rather than revenue, stress churn and acquisition cost, and find the funding source and the rule on which the margin depends. In money-equation terms, this is the third factor written as arithmetic, reaching customers for less than they are worth, and it fails fastest when the second factor, risk pricing, is assumed rather than tested. For the wider picture see our guide to how financial companies make money and our comparison of neobanks and traditional banks, and read Chapter 15, "Business Models and Unit Economics" for the full method.