The State-Level Map That Changes How You Evaluate Indian Fintech

India processed 21.7 billion UPI transactions in January 2026. That is a real number and an extraordinary one. UPI is the largest retail payment system in the world by volume, settling more transactions in a single month than the card networks of many large economies process in a year.

If you are investing in Indian fintech, that number is the first thing you will hear in every pitch. It is also the wrong number to start with.

The gap the headline hides

I first looked at state-level digital payments for my undergrad thesis in 2024, covering the early-UPI era (FY 2019-21) when UPI was one of several rails competing for volume. A few weeks ago I went back and extended the analysis to the mature era (2023-26) to see how the state-by-state story had shifted. The methodology is the same across both: pooled OLS with year fixed effects and state-clustered standard errors, same controls, same log-log specification. The fact that the same analysis was run on two different eras, five years apart, is the point. What changed and what stayed the same is visible in the results.

The national average for UPI transactions is about 8.3 per person per month, and underneath that average is a sixfold gap. Telangana and Goa sit near 22. Bihar and Tripura sit near 3.5. The highest-use states do roughly six times the per-capita volume of the lowest-use states.

The income-tier coloring in the chart below makes the pattern hard to miss. The top of the distribution is almost entirely high-income states (dark blue). The bottom is almost entirely low-income states (red). Income is doing most of the work. The regression confirms it: per-capita state income has an elasticity of roughly +1 in both eras. A 10 percent richer state uses about 10 percent more digital payments. That relationship has not changed since 2019.

The gap is closing, slowly. The chart below tracks two measures of cross-state dispersion over 34 months, both indexed to their April 2023 starting value. Both decline steadily. Low-use states are growing their per-capita UPI faster than high-use states.

But convergence is not parity. The starting point was roughly a tenfold gap between the highest and lowest-use states; it has narrowed to about sixfold. Anyone underwriting a fintech on the assumption that "India is converging" needs to know the convergence is real, the distance remaining is large, and the pace is measurable but not fast.

Why this changes fintech evaluation

Most fintech pitch decks present growth at the national level. Monthly active users, transaction volume, GMV, all reported as a single number. The implicit claim is that the company is riding the India wave. The national numbers make that wave look enormous. It is enormous. It is also deeply uneven, and the unevenness is where most of the interesting questions are.

Here is what the state-level data changes about evaluation, structured around the three questions I would want answered before backing a fintech betting on tier-2 India.

1. Where is the growth actually being harvested?

This is the question that almost never gets asked, and it should be the first one.

A fintech posting 40 percent user growth tells you almost nothing until you know which states that growth is coming from. Pune sits in Maharashtra, a state at 16.6 per-capita monthly UPI transactions. Lucknow sits in Uttar Pradesh, a state at 4.5. A company growing in Pune is adding users in a market that is already near the top of the national range. The same growth rate in Lucknow is adding users in a state with structural room to triple or quadruple.

Same number on the slide. Opposite trajectory underneath it. One company is running out of road. The other is just getting on it. The blended national figure cannot tell you which is which. The state breakdown can.

The practical ask: any fintech operating in India should be able to show you its user growth disaggregated by state, or at minimum by state tier. If it cannot, the national growth number is uninterpretable.

2. Is the growth riding income, or going beyond it?

The single strongest finding from the analysis is how much income explains. Across 36 states, in two separate time periods, with different data and different dependent variables, income alone accounts for most of the cross-state variation in digital payment use. The elasticity is close to one in both eras. It has not moved.

This matters for a fintech investor because it means that most of the "growth" in digital payments across Indian states is just income growth expressed through a fintech product. A state gets richer, its residents use more UPI, and any fintech operating there gets lifted by the tide. That is not traction. That is macroeconomic correlation.

The interesting question is whether a fintech has found growth beyond the income gradient. This is harder to see, but it is the more important question. The analysis finds one variable that does add explanatory power beyond income: PMJDY, the government's flagship financial-inclusion program that opened bank accounts for the unbanked. States with broader PMJDY enrollment use more digital payments than their income alone predicts, and that finding holds in both eras. The mechanism is account access. PMJDY deliberately targets people outside the income gradient, and that is what gives it separation from income in the data.

For an investor, the read is: a fintech whose growth tracks the income gradient is riding the macro wave. A fintech whose growth correlates with inclusion indicators beyond income may be doing something structurally different. The former decelerates when GDP growth slows. The latter may not.

There is a counterintuitive turn in the PMJDY data that I will write about separately.

3. Is the ground ready?

Not every low-use state is an opportunity. Some are frontier markets with real room to grow. Others are low-use because the infrastructure needed for digital payments is not in place, and no amount of product quality will fix that on a fintech's timeline.

The data shows you which is which. Internet density, bank-office coverage, and urban share all correlate tightly with state income (correlations of +0.85, +0.67, and +0.7 respectively). In high-income states, these infrastructure variables are not binding. In low-income states, they are genuinely constraining. A fintech trying to scale UPI adoption in a state where a large share of the population does not have reliable internet access is not entering a frontier market. It is entering a market that is not yet ready for its product.

The practical version of this question: what is the digital access floor in the target state, and does the company's growth plan account for it or assume it away? A strong product in a state that is not ready is an expensive lesson. That lesson is avoidable if you look at the state-level infrastructure data before you look at the pitch deck.

A related nuance: even within above-average states, the city-level picture can diverge from the state average. Pune is one of the richest cities in Maharashtra, which is itself near the top of the national range. A fintech "in Maharashtra" might really be "in Pune," which is a near-saturated micro-market inside a near-saturated state. The state number is a better evaluation altitude than the national number, but it is still not the finish line. The investor should always ask whether traction is in the state's cities or in the state broadly. Urban-only traction in an above-average state is the weakest version of "tier-2 expansion."

The meta-question

There is a bigger question underneath all three, and it is about the company's thesis, not its metrics.

Is this company's pitch "India is big" or "India is uneven"?

The "India is big" pitch starts from the national headline. 21.7 billion transactions, largest in the world, massive TAM. It is true. It is also the pitch that every fintech in the country can make, because it describes the country, not the company.

The "India is uneven" pitch starts from the state map. It says: here is which states we are in, here is how much room each of those states has, here is what the access floor looks like in each of them, here is why our growth in these specific states is not just income riding up. It is a harder pitch to make because it requires knowing the answers to the questions above. It is also a more defensible one, because a company that knows which India it is operating in is a company that has done the work.

The state-level data does not tell you which company to back. It tells you which questions to ask and whether the answers you are getting are real. For a market this large and this uneven, that is probably the most useful thing a dataset can do.

About the analysis

Data sources: NPCI state-wise UPI volumes, MoHFW population projections, RBI bank-office counts, TRAI internet density, and PMJDY enrollment snapshots from the Department of Financial Services. The full paper is in progress. If any of the findings or methodology would be useful to your work, I am happy to discuss.