Underwrite distribution economics, not transaction economics

The recommendation

Conditional no on a payments-fee thesis; conditional yes on a merchant-financing thesis. The merchant payments market is enormous and still compounding, but at 0bps MDR the payment itself is not a revenue event. An investment here has to be underwritten as distribution economics: the right to lend to, and sell software to, a merchant base, not as transaction economics. If the sponsor’s model depends on MDR returning, the answer is no, because that is a policy bet, not a business plan.

What the numbers say

Metric (2026Q2)	Value
Registered merchants	50,705,249
Transactions per merchant per quarter	487
GMV per merchant per quarter	Rs 206,685
Merchant GMV	10.48 lakh crore
Average merchant ticket	Rs 425
Payment revenue at today’s 0bps MDR	Rs 0

Annualised revenue per merchant, if an MDR existed:

MDR	Revenue per merchant per year	Total
10bps	Rs 827	Rs 4,192 crore
30bps	Rs 2,480	Rs 12,576 crore
50bps	Rs 4,134	Rs 20,960 crore

Three supporting arguments

1. Market attractiveness is not the constraint; price is. 50,705,249 merchants each running 487 transactions a quarter is a distribution asset most lenders would pay a great deal to rent. The constraint is that the transaction carries no price.

2. The unit economics only work at the merchant level, and only with a second product. At 30bps: a rate that does not exist: a merchant is worth Rs 2,480 a year. Any working-capital product priced off the same relationship dwarfs that. The payment is the acquisition channel; the loan is the P&L.

3. The only listed pure-play has not solved it yet. One97 Communications (Paytm) reported Rs 8,437 crore of revenue in FY2026 at a 7% net margin. A sponsor cannot underwrite this deal on a ‘payments scale economics’ thesis when the largest listed comparable, with a decade of scale, still earns a thin return on the payments business itself.

Red flags

• Policy dependency. Any model that assumes MDR returns is underwriting a political decision. Size the downside at 0bps forever.
• Concentration. The top players intermediate the large majority of volume; a binding 30% share cap redistributes rather than grows the pool, and could force uneconomic customer acquisition.
• Credit is a different business. The pivot to lending swaps a capital-light fee model for a balance-sheet, provisioning and collections business. Underwrite the team for the business they are becoming, not the one they built.
• Take-rate opacity. Reported “payments revenue” at Indian platforms blends MDR-bearing instruments, PPI interchange and incentives. Insist on a take-rate bridge by instrument before signing.

Upside register

• Credit on UPI at scale, where interchange is permitted, converts the base into a fee-bearing channel without a policy change.
• A tiered MDR exempting small merchants would monetise large-merchant GMV with political cover. Low probability, high impact: a genuine option, not a base case.
• Merchant software and settlement, priced as SaaS, is chargeable today and is not exposed to the MDR debate at all.

Ambiguity register

The JD asks for the ability to deal with ambiguity and to develop approaches to tackle diligence questions. Stated explicitly, here is what this analysis does not know, and how a real diligence would resolve it:

Unknown	Why it is unresolved here	How to resolve it
True market-wide merchant count	Pulse discloses one operator’s registered merchants; merchants multi-home across apps	Acquirer-level data room; NPCI acquirer reporting under NDA
Actual blended take rate	Not disclosed at instrument level in any open source	Management take-rate bridge; sample settlement files
CAC and payback by merchant segment	Not observable externally at all	Cohort files from the target; channel-level spend
Credit loss on UPI-originated lending	Vintages too short and not public	Static-pool loss curves by vintage from the lending partner
Whether the 30% share cap binds	The enforcement date has moved before	Regulatory counsel; model the cap as a step function

How this changes the recommendation. The conditional no is robust to all five: none of them makes a 0bps payment fee-bearing. The conditional yes is fragile to the credit-loss question specifically: if UPI-originated loss rates run materially above unsecured norms, the merchant-financing thesis fails and there is no third leg to fall back on. That single unknown is where diligence spend should concentrate.

Scope note

This is a simulation built entirely on public data, to demonstrate diligence structure. It is not investment advice and not a recommendation on any security. One97 Communications appears only as the listed comparable, using its filed figures.