Where a fintech payment actually earns
Moving money is nearly free at the margin, so payments businesses rarely charge the sender. They earn a small percentage from the merchant, called the merchant discount rate, plus fees for instant settlement or added services.
A platform processing 5,00,000,000 a month at a 0.35% merchant rate earns 17,50,000. Handling 4,00,000 transactions at 2.5 each costs 10,00,000, leaving 7,50,000 before fixed costs and fraud losses.
The business is therefore a volume business with thin margins. Scale lowers the cost per transaction, and fraud losses can erase the margin entirely if controls are weak.
Key points
- Merchants, not senders, usually pay for payments.
- Margins are thin, so volume and cost per transaction decide profitability.
- Fraud is a direct deduction from a thin margin.
Unit economics of a digital loan
Contribution = Revenue − Expected loss − Servicing cost
What each loan adds before fixed costs.
Break-even loans = Fixed costs ÷ Contribution per loan
The number of loans needed to cover the platform.
Each loan brings 1,800 in interest and fees over its life. Expected credit loss is 500 and servicing costs 300, leaving a contribution of 1,000 per loan.
With 25,00,000 of fixed platform cost each month, the business breaks even at 2,500 loans a month. Below that it loses money on every additional loan only in the sense that it has not yet covered the base; above it, each loan adds 1,000 of profit.
The tempting shortcut is to grow the loan book faster than the loss estimate. When underwriting is relaxed to hit volume, the 500 loss figure quietly rises and the break-even point moves out of reach.
Key points
- Contribution per loan, not revenue per loan, funds the platform.
- Break-even volume depends on an estimate of losses that can be wrong.
- Growth bought with looser underwriting is the classic failure mode.
Paying for a customer
Payback = Customer acquisition cost ÷ Contribution per loan
How many loans it takes to earn back the cost of winning a customer.
If winning a customer costs 1,200 and each loan contributes 1,000, the second loan repays the acquisition cost. A customer who borrows twice is worth 2,000 in contribution against 1,200 spent, a ratio of 1.67.
Repeat borrowing is what makes digital lending work, which is why so much of the product is designed around returning customers rather than new ones.
A ratio barely above one is fragile. A small rise in loss rates or acquisition costs turns a healthy model into a loss-making one.
Key points
- Payback is measured in loans, not months, when revenue is per-loan.
- Repeat use carries the economics of digital lending.
- A thin lifetime-value ratio has no room for error.
Why speed changes the risk
An instant loan is approved in seconds from data the customer already has: transaction history, device signals, and account behaviour. That is faster and cheaper than a human underwriter, and it is also trained on the past.
Models that perform well in stable conditions can fail sharply when the economy turns, because the relationship they learned no longer holds. Speed magnifies this: a bad model can scale losses far faster than a branch network could.
Real-time payments add a different risk. Once a transfer settles, it cannot be reversed, so fraud must be caught before the money moves rather than recovered after.
Key points
- Automated underwriting is fast but learns from history.
- Speed scales losses as readily as it scales lending.
- Irreversible payments need prevention, not recovery.
02 / Calculate
Find the break-even point.
Set fixed costs, price, and variable cost. The chart shows where revenue crosses total cost.
Fixed costs of ৳90,000 a month, a ৳180 average bill, and ৳70 variable cost per order.
Contribution margin
৳110.00
CM ratio
61.1%
Break-even revenue
৳1.47 L
Break-even volume
818units
Each unit contributes ৳110.00. You need 818 units to cover ৳90,000.00 of fixed cost.
Profit at expected units
৳42,000.00
Margin of safety
31.8%
Revenue against total cost
Where the gold revenue line crosses the navy cost line is the break-even volume.
This model assumes one product, a constant price, and a constant variable cost per unit. Real demand falls as price rises.
03 / Apply
A wallet chases growth
A wallet offers instant micro-loans to double its active users. Approval rates rise, marketing spend rises, and for two quarters the losses look acceptable. Then a slow season arrives and repayment rates fall.
The assumptions
- Contribution per loan is positive at the launch assumptions.
- Acquisition cost is paid up front and recovered over repeat loans.
- Loss rates are estimated from a period of stable incomes.
What to take away
01
Volume does not fix a negative contribution per loan.
02
Loss estimates made in good times are optimistic in bad ones.
03
The break-even point moves when the loss assumption moves.
Your turn
0 of 4 answered
01Question
A loan earns ৳1,800, expected loss is ৳500, and servicing costs ৳300. What is the contribution per loan?
02Question
Fixed platform costs are ৳25,00,000 a month and contribution per loan is ৳1,000. How many loans are needed to break even each month?
03Question
A platform processes ৳5,00,000,000 at a 0.35% merchant rate and incurs ৳10,00,000 of transaction cost. What is the net monthly revenue?
04Question
Why is fraud more dangerous in instant payments than in card payments?
Take it into the Lab
Experiment with the same idea.
The Lab turns this topic into a workspace: change the drivers, watch the statements move, and see the consequence before you commit to a decision.
What-if analysis
Change price, cost, and working capital. Watch margins, return, and risk move together.
Open the labFormula lab
Understand, manipulate, observe, solve, then apply the formula to a decision.
Walk the formulaStatement playground
See how one driver propagates through the income statement, balance sheet, and cash flow.
Open the playgroundSources & further reading
Bank for International Settlements · Fast payments ↗OpenStax · Principles of Finance (free textbook) ↗Educational sample content. Worked examples use hypothetical figures. Dates refer to the note's latest content update.