The Invisible Category

Two numbers from TRAI's 2024–25 performance indicators sit awkwardly beside each other.

Industry adjusted gross revenue reached ₹3,72,097 crore in FY25, up 10.72% on the year. Over the same twelve months, total wireless subscribers fell from 1,165.49 million to 1,156.99 million — a net loss of 8.5 million people, and the first absolute decline in years.

More money. Fewer customers. The category has largely filed this under successful tariff repair.

I think it is the most interesting thing that has happened in Indian telecom in a decade, and the industry is reading it backwards.

What the ARPU inversion actually says

Industry ARPU rose 16.89% to ₹174.46. Underneath that average, something stranger happened. Prepaid ARPU climbed from ₹146.37 to ₹173.84. Postpaid ARPU fell, from ₹184.63 to ₹180.86. By Q1 FY26, prepaid ARPU of ₹187 had passed postpaid at ₹182.72 — the first time in the history of the Indian market. The gap between the two, which stood at roughly ₹160 in 2020, had closed to about ₹5 by mid-2025.

Every retention model in Indian telecom treats plan type as a proxy for customer value. Postpaid is the high-value cohort. Prepaid is the volume base you defend on price. That assumption is now empirically dead, and almost no operator has adjusted for it.

Here is what the inversion means in plain terms. When the 2024 tariff hikes landed, prepaid customers absorbed a roughly 19% increase and stayed. Postpaid customers, on higher-commitment contracts, did not deliver a corresponding rise — their ARPU went backwards. The cohort that proved most willing to pay more was the one the industry had classified as most price-sensitive.

You cannot explain that with a price model. A consumer who accepts a fifth more on their monthly bill and does not move is not optimising cost. They are doing something else, and the industry has no metric for it.

Churn is not a pricing failure

TRAI's mobile number portability data shows cumulative requests reaching 1,258.77 million by December 2025, with monthly porting running at 13 to 15 million and holding at record levels since July 2025.

At the industry average ARPU, that monthly churn represents somewhere around ₹27,000 to ₹31,000 crore of annualised revenue at risk — roughly 7 to 8% of total industry AGR. That is not a customer-experience line item. It is a capital efficiency problem large enough to show up in a valuation.

And the category's answer to it is structurally strange. Acquisition is price-led, which by construction recruits the customers with the weakest attachment and the highest propensity to leave again. Retention is discount-led, deployed at the moment the customer has already decided. Upgrade is feature-led, offering larger data buckets to people who will use them exactly as before.

Three interventions, all aimed at the same variable. None of them aimed at whatever it was that made a prepaid customer swallow a 19% increase without moving.

The comparison worth making

T-Mobile US reported 903,000 postpaid net additions in Q4 2024, with more than 60% of new customers choosing its premium Go5G tiers — plans that bundle Netflix and Apple TV+. It then guided to 5.5–6.0 million net additions for FY25 against a consensus of 2.7 million.

T-Mobile did not win that on network speed and it did not win it on price. It won it by making the plan itself mean something to the person holding it.

India's headroom here is enormous and mostly unclaimed. The leading premium operator reported ARPU of around ₹256 in September 2025, the disruptor around ₹211, the merged entity around ₹167. Even at the top, Indian ARPU is a fraction of global comparators — US operators average roughly ₹4,300 a month, European operators around ₹1,000, Chinese operators around ₹600. Projections have Indian industry ARPU reaching ₹220 by end-FY26 and possibly ₹300 by FY27.

Whether that trajectory holds will not be settled by the next tariff cycle. Tariff hikes work once. They do not compound.

Where this argument stops

It would be dishonest to claim every Indian telecom customer is making an identity decision. At the base of the market they are plainly not. Rural tele-density stands at 59.06% against 131.45% in urban India. In those markets, multiple SIMs, signal-led switching and commodity recharge are rational and observable, and the utility framing describes them accurately.

The error is extrapolating that logic upward. Because base-of-pyramid behaviour dominates the aggregate data, it dominates the models — and the models then get applied to a mid-tier customer who is making a different decision entirely.

Churn is not uniform either. TRAI's porting data shows Zone-I, covering northern and western India, accounting for 52% of requests against Zone-II's 48%, with Uttar Pradesh East, Maharashtra and Madhya Pradesh consistently at the top. A single national retention programme will under-perform in every circle by trying to work in all of them.

The question for the next board meeting

If plan type no longer predicts customer value, what does?

No Indian operator currently measures it. The customer who took a 19% increase and stayed, and the customer on a higher-value contract whose ARPU slid, are recorded in the same systems as though the second is worth more. One of those two is compounding. The other is merely undecided.

Until that distinction is measurable, the category will keep spending its retention budget on the customers least likely to stay, and reading the resulting churn as evidence that prices are too high.