Modeling LTV when revenue depends on deposits, spend, or credit usage

Two people open the same neobank account on the same Tuesday. One moves their salary across, leaves about £2,000 sitting in the account and puts the weekly shop on the card. The other funds £20, buys a coffee, forgets the app. A subscription business would book both at the same price. Here the first is worth perhaps fifty times the second, and nothing in the signup record tells you which one you just paid for.

That is the modelling problem: putting a revenue number on a customer when revenue accrues from what they do with their money rather than from a price they agreed to. LTVLTVLifetime Value: the total revenue (or profit) a customer generates throughout their entire relationship with your business.View full definition → (lifetime value) is still the total net revenue expected from a customer over the relationship. In SaaS it is a price multiplied by a survival curve. In fintech it is a behaviour multiplied by a rate that someone else sets, minus losses that arrive a year after the revenue.

This lesson builds LTV formulas for four fintech revenue shapes: interchange, interest spread, subscription, and credit or merchant fees. Then it shows why the curve is a wave rather than a staircase.

Why fintech revenue is behavioral, not contractual

Most consumer fintech products are free or near-free to open. Revenue comes from usage:

• Interchange fee: the fee (paid by the merchant's bank to the cardholder's bank) that an issuer earns every time a customer uses the card.
• Interest spread: the difference between what a fintech earns on customer deposits or lends out, and what it pays customers. Called net interest margin at the balance-sheet level, but for modelling we care about the spread per active dollar, not the accounting ratio.
• Subscription: a flat monthly or annual fee for a premium tier (a metal card, higher cashback, advanced budgeting tools).
• Credit and merchant fees: interest and fees on balances the customer carries, plus a merchant discount the retailer pays. Affirm's 0% APR offers cost the shopper nothing; the merchant buys the conversion.

A customer who signs up but never funds, never swipes and never borrows generates close to zero, however healthy they look in a dashboard. Which post-funding behaviours actually predict that a customer keeps transacting is the retention lesson's territory; this one assumes those signals and prices them.

Formula 1: interchange-driven LTV

LTV = (Average monthly spend × Interchange rate × Gross marginGross marginGross margin is the share of revenue left after subtracting the direct cost of producing goods or services, expressed as a percentage of revenue.View full definition →) × Average lifetime in months

Worked example:

• Average monthly card spend: $1,200 (realistic for a primary spending account, estimate)
• Interchange rate captured: 1.5% (estimate, varies by network and card type)
• Gross margin after processing costs: 70%
• Average customer lifetime: 30 months (estimate for a mid-retention neobank)

Monthly revenue per user = $1,200 × 1.5% × 70% = $12.60

LTV = $12.60 × 30 = $378

Two things break this model faster than churn does. First, the rate is set by regulators, not by you. In the EU, the Interchange Fee Regulation caps consumer interchange at 0.2% on debit and 0.3% on credit, so a European interchange LTV built on US assumptions can overstate revenue several times over. In the US, the Durbin amendment caps debit interchange at 21 cents plus 0.05% for issuers with $10bn or more in assets, while smaller issuers are exempt and earn multiples of that. Identical customer, identical spend, different economics depending on which sponsor bank the programme sits behind. That is a card-programme decision with a direct line into the marketing model.

Second, rewards come straight off the top. A 1% cashback offer against 1.5% gross interchange leaves you a third of the headline. And if average spend falls from $1,200 to $600 (the usual pattern when a card becomes a secondary account) the LTV halves without a single cancellation.

Formula 2: interest spread-driven LTV

LTV = (Average deposit balance × Spread captured × Gross margin) × Average lifetime in years

Worked example:

• Average balance: $2,500 (estimate for a digital-first checking or savings product)
• Spread captured: 2.5% annually (estimate; moves with central bank rates, so not stable year to year)
• Margin after funding costs and reserve requirements: 80%
• Average lifetime: 3 years

Annual revenue per user = $2,500 × 2.5% × 80% = $50

LTV = $50 × 3 = $150

This LTV moves with central bank policy, not with anything marketing controls. Starling Bank is the clean illustration: it reached its first full year of profit in the year to March 2022 and was posting pre-tax profits in the hundreds of millions two years later, over a period when the Bank of England base rate went from 0.1% to 5.25%. The customers did not change their behaviour. The per-pound value of their balances did. Run the model in reverse and a 200 basis point cut removes a third or more of deposit LTV overnight, which is why deposit-led fintechs should recompute payback assumptions on every rate decision.

The second trap is the mean. Deposit balances are heavily skewed: a small share of customers holds most of the money, so an average-balance LTV describes almost nobody. Model balance deciles separately, or you will price acquisition for a customer who does not exist. Deposits are also callable in a day. A contract has notice periods; a balance does not.

Formula 3: subscription-driven LTV

The closest fintech gets to SaaS logic, used by premium neobank tiers and flat-fee robo-advisors.

LTV = Monthly subscription fee × Gross margin × Average customer lifetime (months)

Fee $9.99, gross margin 85%, lifetime 24 months gives ~$204. Stable and forecastable. Pure subscription fintechs are rare, so real models are usually hybrid, summing two or three lines per segment.

Formula 4: credit and merchant-fee LTV, net of losses

LTV = (Balance carried × Net interest margin + Fees + Merchant discount) × Lifetime − Expected credit losses − Cost to serve

Affirm earns on two lines: a merchant discount on interest-free instalment plans and consumer interest on interest-bearing loans. It has guided to revenue less transaction costs of roughly 3 to 4% of GMV, and reports around five transactions per active consumer per year. Put those together: a $250 average basket, five purchases, 3.5% net take gives about $44 a year, so a three-year relationship is worth roughly $130 before fixed costs. Frequency, not basket size, is the lever, and frequency is a merchant-network problem as much as a marketing one.

The failure mode here has no SaaS equivalent: credit LTV can be negative. A cohort acquired at the edge of the credit box books revenue in months 1 to 6 and charges off in months 9 to 18, long after the growth team has declared payback. Nubank grew on a credit card in a market with high loss rates, which is why its unit economics are always quoted risk-adjusted rather than gross. If your model does not carry a provision line by vintage, it is a revenue model wearing an LTV label.

Why the LTV curve shape differs from SaaS

A SaaS payback model assumes flat monthly revenue until cancellation: a staircase that steps to zero. A fintech curve is a wave:

1. Ramp-up: new users take weeks to move a salary, set up bill pay or fund a balance. Revenue starts near zero.
2. Peak engagement: revenue rises as the product becomes primary.
3. Decay or plateau: spend and balances shrink as customers spread across multiple apps (most consumers hold several, per surveys from firms like Plaid, which sells the connectivity layer those apps run on).
4. Long tail: some users keep a small balance or occasional spend indefinitely, an ambiguous zombie state with no cancellation event to date.

So plot revenue-weighted cohort curves, not survival curves: spend and balance per surviving customer by month since funding. A cohort with 90% logo retention and half the spend per head is a cohort that lost half its value while looking healthy.

Putting it together: a simple blended model

def blended_ltv(monthly_spend, interchange_rate, avg_balance, spread_rate,
                 sub_fee, margin, lifetime_months, monthly_credit_loss=0.0):
    interchange_rev = monthly_spend * interchange_rate * margin
    spread_rev = (avg_balance * spread_rate / 12) * margin
    sub_rev = sub_fee * margin
    monthly_total = interchange_rev + spread_rev + sub_rev - monthly_credit_loss
    return monthly_total * lifetime_months

# Example: hybrid neobank customer
ltv = blended_ltv(
    monthly_spend=1000, interchange_rate=0.015,
    avg_balance=1800, spread_rate=0.02,
    sub_fee=4.99, margin=0.75,
    lifetime_months=28, monthly_credit_loss=1.50
)
print(round(ltv, 2))  # illustrative output only

Growth and finance build this jointly. The output is not one number; it is a distribution across cohorts, by channel, by balance decile, by whether a salary landed in the first 30 days.

Anchors for the revenue side (rules are hard, ratios are estimates)

• Interchange caps are law, not benchmarks: 0.2% debit and 0.3% credit in the EU, 21 cents plus 0.05% for large US debit issuers. Check which regime your card sits in before trusting any interchange LTV.
• Affirm's stated target of 3 to 4% revenue less transaction costs on GMV is a usable ceiling for merchant-fee models in instalment credit.
• Nubank has reported monthly revenue per active customer of roughly ten dollars against a cost to serve under a dollar, with older cohorts well above the average. Small ARPAC plus tiny serving cost is a viable shape; small ARPAC plus a branch is not.
• Whether the resulting LTV is healthy against your CACCACCustomer Acquisition Cost (CAC) is the total sales and marketing spend divided by the number of new customers gained in a period. It measures how efficiently you grow.View full definition → is the benchmarking lesson's call, and the answer differs by sub-category.

Public, audited fintech marketing benchmarks are scarce. Most circulating figures come from investor decks, conference talks or aggregations like a16z's fintech benchmarks, published by a firm that invests in the companies reporting them.

🎬 [VIDEO: "How Neobanks Make Money" - youtube.com - a walkthrough of interchange, interest spread, and subscription revenue models used by digital banks]

Key Takeaways

• Fintech LTV is behavioral, so model it per mechanism: interchange, spread, subscription, credit and merchant fees, usually blended per segment.
• Interchange LTV is capped by regulation before it is shaped by marketing (0.2/0.3% in the EU, Durbin limits in the US), and cashback comes off the top.
• Deposit LTV moves with central bank rates, as Starling's swing through the 0.1% to 5.25% cycle shows, and average balances hide a skew that makes decile modelling necessary.
• Credit and merchant-fee LTV must net out expected losses by vintage; a cohort can look profitable for a year and finish negative.
• Track revenue per surviving customer by cohort month, not just retention, because value decays long before anyone closes an account.