Modeling customer lifetime value for deposit and card holders

Your head of deposits wants to raise the checking signup bonus from $250 to $400, and wants an answer this week. There is one honest way to reply: build what the customer is worth, then compare. That worth is not a single revenue line. A deposit and card customer pays you through interchange or a merchant discount rate, through the spread on the balance they leave sitting with you, through interest if they revolve, and through whatever fee income survives. This lesson builds each line, discounts it over expected tenure, and then stress-tests the answer.

The revenue streams of a card and deposit customer

1. Interchange and the merchant discount rate

Interchange is the fee a merchant's bank pays the card-issuing bank on every swipe. The customer never sees it. On a US debit card it is capped by the Durbin Amendment (part of the 2010 Dodd-Frank Act) for banks with over $10 billion in assets, at roughly 21 cents plus 0.05 percent of the transaction. Smaller banks and credit unions sit outside the cap, which is why community banks and neobanks riding a small sponsor bank can afford louder debit rewards: their take per swipe is higher.

In Europe the squeeze is harder. The EU Interchange Fee Regulation (IFR, 2015) caps debit interchange at 0.2 percent of transaction value and consumer credit at 0.3 percent. The same customer behaviour produces a fraction of the US revenue line.

Closed-loop networks change the arithmetic again. American Express is both network and issuer, so it collects a merchant discount rate rather than a share of interchange, and that rate has run a little above 2 percent of billed business in recent years, several times what a Durbin-capped debit swipe yields. A model borrowed from an Amex-style P&L and dropped onto a large-bank debit book will be wrong by a multiple, not by a rounding error.

Worked estimate (US, large bank, debit):

• Customer spends $1,500/month on the card.
• Effective interchange yield roughly 0.5 percent (the fixed cap dominates on small tickets).
• Annual interchange = $1,500 x 12 x 0.005 = $90/year.

2. Deposit spread

The customer leaves money in the account, and treasury credits the retail book for it through funds transfer pricingtransfer pricingThe prices charged when units of the same company trade goods, services or IP across borders, set to satisfy tax authorities and reflect market value.View full definition →: the internal rate the bank would otherwise pay for that funding, minus what you pay the customer. In a 5 percent policy-rate world a near-zero-rate checking balance shows a gross spread near 4 points, but marketing rarely gets credited the gross number. Liquidity charges, reserve treatment and expected deposit beta take a haircut off it first.

Worked estimate:

• Average balance $2,500.
• Marketing-credited deposit spread: 2 percent (a treasury assumption, not a published figure).
• Annual deposit contribution = $2,500 x 0.02 = $50/year.

Two things break this line. First, rates move and your customer does not: a customer acquired at a 4 point gross spread is worth materially less two years later after cuts, with identical behaviour. Second, balances are skewed. A cohort mean of $2,500 often hides a median nearer $800, so modeling the average customer credits most of the cohort with money it never holds. Run the deposit line by balance decile or you will overpay for the bottom half. Stability is the other half of the question: DBS has consistently pointed investors to its current and savings account base as the reason its margins held, and its 2017 disclosure that digitally engaged consumer customers generated roughly twice the income of traditional ones is the same LTVLTVLifetime Value: the total revenue (or profit) a customer generates throughout their entire relationship with your business.View full definition → arithmetic run across these lines.

3. Revolving interest, for card holders

A credit card customer who carries a balance is a different animal. US card APRs on accounts assessed interest have sat above 20 percent, so a $2,000 average revolving balance generates hundreds of dollars a year, an order of magnitude past interchange. It also brings credit losses; industry card charge-offs run in the low-to-mid single digits of balances, and they arrive late in the relationship, after your campaign has been declared a success.

Do not average transactors and revolvers into one customer. They have different revenue, different loss curves and different attrition, and the mix shifts with the acquisition channel. A rewards-led premium acquisition skews to transactors and lives on discount rate plus annual fees, which is roughly the American Express shape; a balance-transfer offer buys revolvers and lives on interest net of losses.

4. Residual fee income

Overdraft, out-of-network ATM and wire fees still exist, though US banks have cut overdraft pricing sharply since 2022 under CFPB pressure. Model this conservatively and assume it keeps shrinking: $15/year per active customer.

Putting annual revenue together

Subtract annual cost to serve (fraud, rewards paid out, app and support). Assume $55/year.

Annual contribution margin = $155 - $55 = $100/year.

Adding expected tenure

LTV needs a horizon, and the input is annual retention, measured against the live-relationship signals the app engagement lesson defines rather than an open account with a zero balance. If retention is 90 percent:

Average tenure (years) = 1 / (1 - retention rate)
                       = 1 / (1 - 0.90)
                       = 10 years

The formula assumes constant hazard, which is where most bank LTV models quietly inflate. Attrition is front-loaded: bonus-chasers satisfy the direct-deposit qualifying period, collect, and close within a couple of months. Applying a mature 90 percent rate to a fresh promo cohort will overstate LTV badly. Fit retention from month 13 onward and model the first year as a separate survival curve.

Computing LTV

LTV = annual_margin x (retention / (1 + discount - retention))

Traditional bank customer, margin $100, retention 0.90, discount 0.10:

LTV = 100 x (0.90 / (1 + 0.10 - 0.90))
    = 100 x 4.5
    = $450

Neobank customer, retention 0.78, same margin and discount:

LTV = 100 x (0.78 / (1 + 0.10 - 0.78))
    = 100 x 2.4375
    = $244

Same annual margin, and higher churn nearly halves lifetime value.

The discount rate deserves an argument, not a default. Ten percent is convention; using the bank's cost of equity is more defensible. Move it to 15 percent and the $450 customer becomes $360, a fifth of the value gone with no change in behaviour. Fix the rate with finance once, in writing, or every campaign business case will pick the one that flatters it.

Reconciling against acquisition costacquisition costCustomer 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 →

Take the funded-account cost per customer the acquisition lesson teaches you to build, bonus included. Say it lands at $370: a $250 checking bonus (US promos have run roughly $200 to $400 in recent cycles) plus $120 of media and channel cost.

LTV:CAC = 450 / 370 = 1.22

The 3:1 rule everyone quotes came out of subscription software, where a vendor like Drift (a conversational marketing platform, so it sells into exactly this debate) renews or loses a contract every twelve months and collects cash upfront. A deposit relationship has neither property. Banking runs lower on the anchor product because of cross-sell, but 1.22 on the debit product alone is a warning unless the cross-sell is real. Whether your number is genuinely good against the market is a comparability question the benchmarks lesson handles.

Cross-sell changes everything

If 25 percent of these customers later add a product worth $600 in incremental LTV:

Blended LTV = 450 + (0.25 x 600) = $600
Blended LTV:CAC = 600 / 370 = 1.62

The failure mode here is that the 25 percent is almost always assumed rather than measured on the acquired cohort. Realise 8 percent instead and blended LTV is $498, ratio 1.35, and the campaign that was approved on a 1.62 was never that. Backtest the cross-sell rate on last year's bonus cohort before it enters this year's model.

Payback period, the metric CFOs actually watch

Monthly margin = 100 / 12 = $8.33
Payback = 370 / 8.33 = ~44 months

Almost four years, against a common internal target of 18 to 24 months. Worse, the bonus is expensed at payout while margin accrues monthly, so a strong acquisition quarter looks like a bad P&L quarter. And a customer who takes the bonus in month 3 and leaves in month 6 never pays anything back in cash. Trim the deposit spread by a fifth after a rate cut and this payback slides past 55 months.

Inputs you do not get to set

Three of the numbers above belong to someone else, and marketing owns none of them: the interchange cap (Durbin in the US, IFR in Europe), the funds transfer price treasury credits you for balances, and the discount rate finance approves. Sensitivity-test all three rather than arguing about them.

What you do own is retention and card spend, and they are not equal levers. Moving retention from 0.85 to 0.90 lifts LTV from $340 to $450. Lifting monthly card spend by 30 percent at the old 0.85 retention gets you to $432. The retention point wins, and it compounds into the cross-sell assumption too.

Key takeaways

• Model four lines separately: interchange or merchant discount rate, deposit spread from funds transfer pricing, revolving interest net of losses, and residual fees. Transactors and revolvers are different customers.
• Average tenure equals 1 / (1 - retention), but attrition is front-loaded. Fit retention from month 13 for bonus cohorts or you will overstate LTV.
• The discount rate is an input, not a constant: 10 to 15 percent moves LTV by a fifth. Agree it with finance once.
• Balance and spend distributions are skewed. Run the model by decile; the cohort mean flatters the bottom half.
• A 44-month payback means a rate cut or a churn surprise can retrospectively make a campaign unprofitable, so pressure-test the bonus before it ships.