What Makes a Neobank Profitable? A Unit-Economics Model

What makes a neobank profitable is not a single product or customer-count threshold. It is the relationship between revenue per active customer, funding and credit risk, acquisition and servicing cost, and the fixed cost of operating a regulated platform. Two digital banks with the same number of registered users can have entirely different economics.

Public examples show that profitable digital banking businesses exist, but company announcements must be read as company-specific evidence. Monzo's 2024 annual report^[1] recorded its first full year of profitability, while Revolut reported record profit in its 2025 results announcement^[2]. Neither result establishes a universal formula for every neobank.

The unit-economics equation

Start with active customers rather than downloads or opened accounts. For each customer cohort, measure interchange retained by the bank, net interest income, lending and fee income, and subscription revenue. Subtract rewards, partner and scheme fees, expected credit loss, fraud and disputes, support, compliance operations, infrastructure and acquisition amortisation.

The model should be built by acquisition month and geography. A blended company average can hide an unprofitable recent cohort, a market with materially different interchange, or a temporary boost from high interest rates.

Interchange depends on jurisdiction and contracts

Consumer interchange is constrained in Europe. The EU's Interchange Fee Regulation^[3] caps consumer debit at 0.2% and consumer credit at 0.3% for in-scope transactions. The UK implemented the same domestic caps through its payment-card regime. That limits the gross issuer revenue available from an otherwise identical card purchase.

The US structure is different. The Federal Reserve's 2024 network data^[4] reports exempt dual-message debit transactions averaging 1.41% of transaction value, while covered dual-message transactions averaged materially less. The Federal Reserve cautions that its averages do not state what a particular issuer will earn on a particular network or programme.

That caveat matters for fintechs using a sponsor bank. Headline interchange is not necessarily the neobank's retained revenue: the bank, network, processor, programme manager and rewards programme may all take part of the economics. A defensible model starts with the contracted net share actually received.

Deposits and lending can dominate card revenue

A funded account creates balance-sheet economics that an unfunded card does not. Deposit income depends on average customer balances, deposit pricing, liquidity requirements and the yield the bank or sponsor can realise on assets. It changes with rates, so a model should stress lower-rate and higher-rate scenarios rather than capitalise one favourable year indefinitely.

Lending can add materially more revenue per customer than debit interchange, but only after funding cost, expected credit loss, collections cost and regulatory capital are included. Gross interest yield is not contribution margin. Compare each origination vintage through delinquency and charge-off rather than judging a young book on early interest income.

Transaction data may improve income verification and affordability assessment, but it does not guarantee better credit outcomes. Our article on AI underwriting explains why speed, approval coverage and predictive performance need to be measured separately.

Subscriptions and ancillary products

Paid plans can diversify revenue and increase engagement, but the relevant measure is net subscription contribution after included benefits. Travel insurance, lounge access, higher savings rates, foreign-exchange allowances and rewards all have direct or contingent costs. Gross subscription receipts overstate value when those benefits are ignored.

Wealth, foreign exchange and business accounts can add fee income, but cross-selling is an observed company outcome rather than an automatic property of a large user base. Track attach rate, contribution and retention effect by product. A feature that improves retention may be valuable even if its direct margin is modest; a product with attractive gross revenue may destroy value if support and compliance costs are high.

Acquisition, activation and servicing

There is no reliable industry-wide customer-acquisition-cost range. Referral, paid search, sponsorship and employer or platform distribution produce different numbers, while companies define an "acquired customer" differently. Use the cost per activated, funded customer who remains active after a defined period, not cost per install or account opening.

Servicing cost also varies by cohort. Fraud reviews, chargebacks, vulnerable-customer support, complaints, KYC remediation and account closures can dominate the cost of an otherwise inexpensive digital account. Automation can reduce routine work, but regulated exceptions still require trained people and auditable decisions.

Cloud-native architecture can improve release speed and elasticity, but it does not prove a lower cost per account. The relevant evidence is the company's own infrastructure, incident, support and engineering cost. The implementation trade-offs are covered in our cloud-core migration playbook.

Scale helps only when marginal economics work

Scale spreads fixed platform, licence and governance costs, but there is no universal "several million users" break-even point. A high-balance, efficiently acquired business cohort may contribute at modest customer count. A large base of dormant or reward-seeking retail accounts can remain expensive.

The decision metric is the contribution curve: does each mature cohort repay acquisition and direct servicing within an acceptable period, and does aggregate contribution cover fixed cost? If contribution deteriorates as the bank grows, because fraud, support, rewards or funding become more expensive, customer count alone does not create operating leverage.

A due-diligence checklist

Investors and operators should reconcile five views: registered versus active users; gross versus retained interchange; deposit balances versus realised net interest income; lending yield versus loss-adjusted margin; and acquisition spend versus retained activated customers. Then stress rates, credit losses, reward cost and partner pricing.

A profitable neobank is ultimately one whose mature customer cohorts generate durable contribution, whose risk costs remain controlled and whose aggregate contribution pays for the regulated operating platform. Interchange, credit, subscriptions and B2B services are inputs to that result: they are not guarantees of it.

Sources & methodology. Regulatory figures are drawn from the EU interchange regime and the US Federal Reserve's 2024 network data. Monzo and Revolut examples use the companies' own public reporting and are identified as company-specific evidence. The cohort equation is a CloudFintech analytical framework; it intentionally supplies no universal CAC, margin or scale benchmark.