Part I — Foundations & Digital Public Infrastructure

Chapter 8 · Fintech Business Models and Unit Economics

Part I has shown you the rails, the licences and the regulators. One question remains before payments can open in full: how does anyone make money on rails that are free? This chapter builds the revenue vocabulary — take rate, MDR, interchange, float, credit spread — then the cost vocabulary — CAC, LTV, contribution margin — and runs the arithmetic that explains why UPI costs nothing at the point of use and what that forces every fintech to become.

📖 Chapter 8 of 35 ⏱ ~60 min read 🎓 Session 8 · CILO 2 Core 65% Stable 35% Volatile 0% 📝 Quiz · 30 MCQs

🎯 Learning Objectives

1. The Revenue Question

Every business model in this book is an assembly of three things you now know: a licence (Chapter 6), a set of stack components it rides (Chapter 5), and a revenue line attached to each activity. The last piece is this chapter. And India gives it a twist most markets lack: the country's dominant payment rail earns nothing at the point of use, which means the largest fintech category — payments — cannot pay for itself directly.

Recall the chain Chapters 1 and 3 built. Zero MDR made UPI a near-free public utility. Utility pricing won the market — 24 billion transactions a month by September 2026 — but it also pulled transaction growth and revenue growth apart. Firms responded by climbing: payments became the hook, and monetisation migrated to whatever sits on top of payments — credit, wealth, insurance, data. Value migrated up the stack, and this chapter shows the exact ledgers that made it happen.

📝
The analyst's opening question — memorise it: for any fintech, ask what does one customer cost, and what does one customer pay? Everything in this chapter is that question made arithmetic.

2. The Four Revenue Lines

2.1 Take rate

The take rate is revenue as a share of the value flowing through: revenue ÷ total payment volume (TPV) for a payments firm, revenue ÷ gross merchandise value (GMV) for a marketplace, revenue ÷ assets under management (AUM) for a wealth platform. It is the single most useful ratio in fintech valuation, because it tells you how much of the flow the firm keeps. A payment aggregator charging a merchant 2 per cent on card volume keeps a fraction of that after paying interchange; a mutual fund distributor's trail fee is an embedded take rate on AUM, measured in basis points a year.

2.2 MDR and interchange

MDR — the merchant discount rate — is the fee a merchant pays to accept an electronic payment, quoted as a percentage of the transaction. It is not one fee but a split: the interchange share goes to the card-issuing bank (which fronts the customer's money and carries fraud risk), the scheme fee to the network, and the remainder to the acquirer or aggregator that signed the merchant. When you hear "payments is a multilateral business", this split is why: no single firm controls both ends of a card payment.

2.3 Float income

Float is money that rests with a firm while in transit or in storage: wallet balances, prepaid instruments, funds awaiting settlement, deposits in a payments bank. Earning interest on that rest — investing it overnight in safe instruments — is float income. It is the quietest revenue line and the most regulated: what a PPI issuer or payments bank may do with customer balances is tightly prescribed. Unearned float on customer money is not a business model; it is a custody duty with a yield attached.

2.4 Credit spread

Lending earns the spread: the interest a borrower pays, minus the cost of funds, minus operating cost, minus expected loss. Expected loss is itself a formula worth memorising: EL = PD × LGD × EAD — probability of default, times the loss given default, times the exposure at default. Credit is the deepest revenue pool in finance — spreads are measured in percentage points, payment take rates in basis points — which is the entire economic logic of value migrating from payments to lending.

3. The Cost Side: Unit Economics

Revenue lines tell you where money comes from. Unit economics asks whether it covers the cost of getting it. Four terms do the work. CAC — customer acquisition cost: marketing and incentive spend divided by customers acquired. Contribution margin — revenue per customer per period minus the variable cost of serving them. LTV — lifetime value: contribution across the whole customer relationship. Payback period — months of contribution needed to recover CAC. The working rule: LTV/CAC of roughly three or better, with payback under a year, is a healthy acquisition engine; below that, growth is being bought, not earned.

Now the arithmetic — the session's workshop in miniature, with round numbers chosen to teach.

📝
Worked example — a ₹10,000 digital loan (6 months at 24 per cent): Interest revenue ≈ ₹1,200 (24% × 10,000 × ½). Funding cost at 10 per cent ≈ ₹500. Acquisition cost ₹300. Expected loss at 4 per cent of exposure ≈ ₹400. Contribution: 1,200 − 500 − 300 − 400 = ₹0. The first loan breaks even at best. The business lives in the second loan and every one after — acquisition already paid, data already priced. Small-ticket credit is a repeat-relationship business wearing a loan product's clothes.
📝
Worked example — a payment aggregator's cross-subsidy: On ₹1 crore of merchant card volume at a 2 per cent merchant fee, the PA bills ₹2,00,000 — most of it flows out as interchange and scheme fees, leaving a thin processing margin. On the same merchant's UPI volume the merchant fee is zero, while the PA still pays technology and support costs. The card margin and the SaaS fees pay for the UPI "freebie" — because the merchant, once acquired, is the customer for lending, payroll and software. Zero-MDR rails turn payments into customer acquisition.
📋 Stable content — Reviewed: October 2026 · Next review: April 2027

4. Why UPI Is Free

The question has a policy answer, an economic answer and a strategic answer, and the full truth needs all three.

The policy answer. Free was chosen. Card MDR had already been capped below 1 per cent by 2019 regulation, and in December 2019 the government used its statutory powers to prescribe zero MDR on UPI and RuPay debit transactions from 1 January 2020. Budget-funded incentive schemes followed, compensating banks in part for the running costs of the network. Free at the point of use is a decade of deliberate decisions, not an accident of technology.

The economic answer. UPI's marginal cost per transaction is a small fraction of a card transaction's — a message between accounts over shared infrastructure, no plastic, no interchange-heavy multilateral split. Cheap is not free, though: servers, fraud systems, settlement and support cost real money at 24 billion transactions a month, and currently banks and the system bear it, softened by the incentive schemes. "Free at the point of use" never meant free to run — the bill arrives elsewhere, which is the whole MDR debate of Chapter 12.

The strategic answer. Free payments are the DPI thesis in one move: price the rail at zero so the entire economy adopts it, then capture value in what the rail enables — data flows for underwriting, distribution for credit and insurance, deposits for banks. It is the same logic as Chapter 4's ₹3 authentication: make the utility nearly free, and the monetisable layers above it grow. And the strategy is not frozen: from October 2026, a small MDR returned to parts of UPI, the state's first step in reopening the free-forever settlement. Chapter 12 has the full fight.

🌎
Real World: When a fintech pitch says "we will monetise payments later", translate it: we are paying CAC in processing costs today, and plan to sell the acquired customer credit, wealth or insurance tomorrow. That can be a sound model — it is most of the industry's model — but it stands or falls on the LTV of the up-stack products, never on the payments themselves.
📋 Stable content — Reviewed: October 2026 · Next review: April 2027

5. Business-Model Archetypes

The archetypes you will meet through the rest of this book, each with its revenue line, cost driver and the one metric that tells you whether it works.

ArchetypeLicence behind itRevenue lineKey metric
Payment aggregatorPA authorisation (RBI)Take rate on non-UPI + SaaS feesProcessing margin; merchant retention
Consumer payments app (TPAP)TPAP approval; partner licencesDistribution income up-stack (lending, broking, insurance)Monthly transacting users; product attach rate
Wallet / PPI issuerPPI authorisationFloat within norms + loading/spending feesAverage balance; float yield
Digital lenderNBFC or bank licenceCredit spread net of expected lossContribution per loan; repeat rate
Wealth platformBroker / distributor registrationTrail on AUM; flat per-order feesAUM growth; revenue bps on AUM
B2B infrastructure / SaaSOften none (software); PA if fundsSubscription + per-API/per-transaction feesNet revenue retention
Neo-banking front-endPartner bank's licenceShare of partner interchange/interest; feesActive users per rupee of partner cost

Archetype archetypes are stable; the specific takes and metrics move with the market — re-checked each review cycle.

Read the table's second column with Chapter 6 in hand and the deep point appears: the licence decides which revenue lines are even available. A TPAP cannot earn float (it holds no funds); a PPI issuer cannot earn credit spread (it may not lend); a pure SaaS firm cannot earn either. Business models are not chosen freely — they are assembled inside licence walls, and Chapter 6's map is the wall plan.

6. Reading a P&L Like an Analyst — Part I Closed

Three metric traps close the toolkit, each a sibling of Chapter 3's adoption traps.

GMV without take rate. "We process ₹50,000 crore a month" says nothing until you ask at what take rate, on which instruments — UPI volume at zero MDR contributes revenue of exactly zero, however impressive the chart. Users without activity. Registered users are a vanity metric; monthly transacting users pay the bills — the dormancy lesson of Chapter 2 in commercial form. Revenue without contribution. Revenue net of nothing flatters any model; the ₹0-contribution loan of Section 3 looks profitable until funding, acquisition and expected loss line up under it. In each trap the cure is the same: pair the headline number with its rate, its activity, or its cost.

6.1 Part I is complete

Eight chapters, one argument. Fintech defined and disciplined (Chapter 1); the system it replaced and its three gaps (Chapter 2); the shove and the hockey stick (Chapter 3); identity at ₹3 (Chapter 4); the stack and its model (Chapter 5); the licences (Chapter 6); the regulators (Chapter 7); and now the money — where revenue comes from, what it costs, and why free rails push everyone up-stack. You hold every tool Part II needs.

Part II opens inside the machine. The number every chapter so far has taken as given — a payment that simply happens — becomes seven moving parts: the app, the PSP banks, the virtual address, the switch, the settlement, the mandate. Chapter 9 asks the question no UPI user has ever had to ask, and answers it: how does the money actually move?

🛠 Hands-On: The Unit Economics Breakdown

This is the session's classroom activity, made self-paced. You will build a mini P&L for two products — a small-ticket loan and a merchant payments bundle — then judge whether either is a business. Round numbers are chosen for arithmetic; the structure is the real thing.

Steps

  1. The loan book. A lender makes ₹10,000, 6-month loans at 24 per cent. Funding costs 10 per cent; acquisition ₹300; expected loss 4 per cent of exposure; servicing ₹60 per loan. Compute contribution per first loan.
  2. The repeat effect. Repeat borrowers (no new acquisition cost, expected loss halves to 2 per cent with data) are 60 per cent of volume. Compute blended contribution per loan.
  3. LTV. If the average customer takes 3 loans, compute LTV, and the LTV/CAC at ₹300 CAC.
  4. The merchant bundle. A merchant does ₹80 lakh UPI volume (revenue: zero) and ₹20 lakh card volume at 2 per cent merchant fee, of which the PA keeps 20 per cent after interchange and scheme. SaaS fee: ₹1,200/month. Monthly tech-plus-support cost per merchant: ₹1,000. Compute monthly contribution.
  5. The verdict. For each product, in under 100 words: viable or not, which single change would most improve it, and which licence makes that change legal or illegal.
View Solution / Walkthrough

Step 1. Revenue ₹1,200; funding ₹500; expected loss ₹400; servicing ₹60; acquisition ₹300. Contribution: 1,200 − 500 − 400 − 60 − 300 = −₹60. First loans lose money — by design of the arithmetic, and often of the market.

Step 2. Repeat loans: 1,200 − 500 − 200 − 60 − 0 = ₹440. Blended (60/40): 0.6 × 440 + 0.4 × (−60) = 264 − 24 = ₹240 per loan. The book is a repeat-lending business; the first loan is CAC.

Step 3. Three loans — say one first, two repeats: −60 + 440 + 440 = ₹820 ≈ LTV. LTV/CAC = 820/300 ≈ 2.7 — close to the healthy threshold of ~3; better retention or loss rates push it over.

Step 4. Card revenue: 2% × ₹20,00,000 = ₹40,000 billed; PA keeps 20 per cent = ₹8,000. SaaS ₹1,200. Revenue ₹9,200; cost ₹1,000; contribution = ₹8,200/month per merchant — but note the exposure: 80 per cent of the volume (UPI) earns nothing, so the card share's health decides the whole bundle.

Step 5. Strong verdicts say: the loan book needs repeat rates (or lower EL via data — the ULI/AA rails of Chapter 18) and is licensed to chase them; the merchant bundle needs card volume or SaaS pricing power, and its illegal fantasy would be charging UPI MDR — which stopped being fantasy in October 2026, the fact Chapter 12 builds on.

Key Takeaways

1

Four revenue lines: take rate on flow, MDR split with interchange to the issuer, float on balances, credit spread net of expected loss (PD × LGD × EAD).

2

Unit economics is one question: what does one customer cost, and what does one customer pay? Contribution, payback and LTV/CAC ≈ 3 are the arithmetic.

3

UPI is free by policy (zero MDR, January 2020), by economics (cheap but not free — banks and incentives bear it), and by strategy (hook for up-stack value).

4

The ₹10,000 loan that contributes ₹0 teaches the industry: first loans are CAC; profit lives in repeat lending on data.

5

The licence decides the revenue lines — a TPAP earns no float, a PPI earns no spread. Business models are assembled inside Chapter 6's walls.

Test Your Understanding

Choose one best answer per question. Select Check Answers to score yourself. An explanation appears after each question once checked. Aim for 24/30 before entering Part II.

1. A firm's "take rate" is defined as:

2. The merchant discount rate (MDR) is paid by:

3. Within an MDR split, the interchange component goes to:

4. Float income arises from:

5. Expected loss on a loan portfolio equals:

6. Before the zero-MDR prescription, Indian regulation had already:

7. Zero MDR on UPI and RuPay debit took effect from:

8. After zero MDR, UPI's running costs were borne mainly by:

9. The strategic logic of free UPI is best stated as:

10. The development of October 2026 relevant to this chapter was:

11. A ₹10,000 loan at 24 per cent for six months earns interest of about:

12. With funding at 10 per cent (₹500), expected loss 4 per cent (₹400), servicing ₹60 and acquisition ₹300, the first loan's contribution is:

13. The business lesson of the break-even loan example is that small-ticket lending is:

14. How does a payment aggregator justify serving a merchant's zero-MDR UPI volume?

15. A consumer UPI app's primary revenue source is:

16. A PPI issuer's core economics rest on:

17. A discount broker's flat ₹20-per-order pricing is a take rate that is:

18. A wealth platform's mutual fund trail income is best described as:

19. Contribution margin is revenue minus:

20. "CAC payback period" measures:

21. The commonly used healthy threshold for LTV/CAC is about:

22. "We process ₹50,000 crore monthly" is incomplete as an investor claim because it omits:

23. The user-metric trap to avoid in fintech P&Ls is:

24. A PA promotes UPI acceptance to merchants despite zero MDR because:

25. A payments bank's principal revenue line is:

26. The phrase "free at the point of use" hides the fact that:

27. Which archetype primarily monetises through subscription and per-API fees?

28. This chapter's one-line explanation of value migration:

29. Part II opens, in Chapter 9, with the question:

30. The chapter's opening question for any fintech P&L is: