Blog · Finance metrics and formulas · Retail banking
Credit card customer lifetime value depends on contribution after funding, rewards, credit losses and servicing. Work through a supplied cardholder cohort, discount its projected contribution, and check the costs and original-customer denominator.
Credit card customer lifetime value is the discounted contribution a cardholder cohort is expected to produce after the costs needed to acquire and serve it. The calculation needs interest, interchange and fees alongside funding, rewards, credit losses and servicing. Card spending or revenue alone cannot answer the profitability question.
Cohort contribution in period t = Interest + Interchange + Fees - Funding - Rewards - Credit losses - Servicing
Net value per original customer = [Sum of period contribution / (1 + discount rate)^t - Initial acquisition cost] / Original customers
The expression assumes the initial acquisition cost occurs at acquisition and each period's contribution occurs at period end. Move the discounting dates if your supplied cash-flow convention differs. This is a management estimate of discounted pre-tax contribution, not a reported accounting asset or a regulatory capital calculation.
Define the unit first. A customer can hold several cards, and a reissued card number does not necessarily represent a new customer. Use a stable, de-identified customer identifier to count relationships consistently, with account identifiers retained for the supporting totals.
The general customer lifetime value calculation explains the wider method. Card portfolios need a more explicit cost bridge because financing balances, rewarding transactions and absorbing credit losses affect the contribution from those relationships.
Use a cardholder cohort file with acquisition dates and original customer counts, plus period-level income, expense and loss summaries. Each summary needs a cohort identifier, period, currency, amount and measurement basis. Keep the approved planning assumptions beside the observed history rather than labeling projections as completed results.
Separate revenue from purchase volume and outstanding balances. The Federal Reserve's 2019 credit-card profitability report presents interest income, interest expense, noninterest income, noninterest expense and provisions for loan losses separately. Those are different economic components; its bank-level returns are not a per-cardholder lifetime-value benchmark.
For your own cost bridge, document what funding charges include, when rewards are recognized, whether recoveries reduce credit losses, and which servicing costs are allocated. Decide where acquisition incentives belong so they are deducted once. A central overhead allocation can be useful, but label the result if it changes the contribution boundary.
Retention assumptions need an owner as well. The OCC's credit card lending handbook overview covers issuer operations and associated risks. In this planning exercise, finance owns the cost convention, credit risk supplies loss assumptions, and portfolio management supplies retention and activity assumptions.
A fictional US issuer acquires 1,000 customers with a total initial acquisition cost of $70,000. Its planning file supplies three years of cohort totals, already allowing for attrition and activity changes. The covered active populations are 1,000 in Year 1, 800 in Year 2 and 600 in Year 3; these counts describe the supplied scenario, not additional multipliers.
The annual discount rate is an assumed 10%. The horizon ends after Year 3 with no terminal value. Call the result a three-year capped lifetime-value estimate, since relationships may continue beyond that point.
| Contribution input | Year 1 (USD) | Year 2 (USD) | Year 3 (USD) |
|---|---|---|---|
| Interest revenue | 180,000 | 152,000 | 120,000 |
| Interchange revenue | 50,000 | 42,000 | 33,000 |
| Fee revenue | 20,000 | 16,000 | 12,000 |
| Total revenue | 250,000 | 210,000 | 165,000 |
| Funding cost | 50,000 | 40,000 | 30,000 |
| Rewards cost | 35,000 | 28,000 | 21,000 |
| Net credit losses | 45,000 | 32,000 | 24,000 |
| Servicing cost | 40,000 | 30,000 | 24,000 |
| Total ongoing cost | 170,000 | 130,000 | 99,000 |
| Cohort contribution | 80,000 | 80,000 | 66,000 |
For Year 1, revenue is 180,000 + 50,000 + 20,000 = $250,000. Costs are 50,000 + 35,000 + 45,000 + 40,000 = $170,000. The difference is $80,000. Apply the same subtraction to the remaining columns before discounting anything.
These are constructed planning inputs. The net credit-loss line is the sole economic loss deduction in this example. An accounting provision is not subtracted again.
With the assumed discount rate, the discount base is 1 + 10% = 1.10. Keep full precision while calculating and round the displayed USD results to cents.
| Step | Calculation | Value (USD) |
|---|---|---|
| Year 1 present value | 80,000 / 1.10^1 | 72,727.27 |
| Year 2 present value | 80,000 / 1.10^2 | 66,115.70 |
| Year 3 present value | 66,000 / 1.10^3 | 49,586.78 |
| Total discounted contribution | Sum of unrounded present values | 188,429.75 |
| Initial acquisition cost | Supplied Year 0 outlay | 70,000.00 |
| Net cohort value | Discounted contribution - acquisition cost | 118,429.75 |
| Value per original customer | Net cohort value / 1,000 | 118.43 |
The denominator remains the original 1,000 customers. Dividing by only the surviving population would spread acquisition costs and early contributions over a selected group and would no longer measure value per acquired customer.
Likewise, do not multiply the cohort totals by survival rates again. Attrition is already represented in the supplied totals. A calculation starting from contribution per active customer would instead need the surviving customer count once in each period.
Across the first table, revenue totals $625,000 and ongoing costs total $399,000. Their difference is $226,000, which also equals 80,000 + 80,000 + 66,000. That proves the undiscounted contribution bridge before the timing adjustment.
The discounted bridge is 188,429.75 - 70,000 = $118,429.75 at displayed precision. Retain the unrounded result when checking that value per original customer multiplied by the original count equals net cohort value. Multiplying a rounded per-customer figure can introduce a small rounding difference.
On real files, reconcile the component amounts to the same cohort and periods in the finance extracts. Document shared-cost allocations and differences between economic loss assumptions and ledger provisions. The ledger-based lifetime value guide explains why historical evidence and future assumptions need separate labels.
Increase Year 2 net credit losses from $32,000 to $42,000, keeping every other input unchanged. The extra $10,000 reduces Year 2 contribution to $70,000. Its present-value effect is 10,000 / 1.10^2 = $8,264.46, or $8.26 per original customer.
Net cohort value becomes $110,165.29 and value per original customer becomes $110.17. The lower estimate comes entirely from the changed loss assumption. No extra churn assumption or funding increase was introduced.
This kind of sensitivity comparison helps decide which assumptions deserve review before changing acquisition budgets. Keep current customer profitability separate from lifetime value, since a period result and a projection answer different questions. Evaluate any claimed lifetime-value statistics against the same cost and horizon definitions.
Bring de-identified cohort income and cost summaries, the original customer counts, and your supplied projection assumptions. Covirage maps the columns, and its tools compute and check the figures; the external AI model explains the result with every figure traced to its rows. Create a free Covirage account and bring a sample of your own data.
See Covirage for retail banking for the supporting industry context.
No. Purchase volume is the amount customers spend with merchants. An issuer's revenue inputs can include interest, interchange and fees under its accounting policies. Do not put total purchase volume into the lifetime-value numerator as though it were income earned by the bank.
State the convention explicitly. You can report value before acquisition cost for comparison with CAC, or deduct acquisition cost to report net value. The worked example deducts it once at acquisition and keeps it separate from ongoing rewards and servicing.
There is no useful single average without a defined population, horizon and cost policy. Compare the same card type and acquisition cohort under matched assumptions. A premium rewards portfolio and a basic revolving-credit portfolio can have different income, retention and cost patterns.
Do not deduct both for the same exposure and period. Choose a consistent economic loss measure for the planning calculation. Explain how realized net losses or supplied expected losses relate to accounting provisions, recoveries and timing before reconciling the result to reported earnings.