How to Analyse a Telecom Stock (ARPU, Churn, AGR)

In the September 2022 quarter, Vodafone Idea's managing director was asked, on the record, why the company kept losing customers. He did not blame competition or pricing. He said the quiet part out loud: "the churn is getting impacted because of the funding not being available today" (Vodafone Idea Q2 FY23 concall). Read that again. The customers were not leaving because rivals were cheaper — they were leaving because the company had no money to build the network that would keep them. The same quarter, Vodafone Idea's subscriber base fell to 234.4 million from 240.4 million three months earlier (Vodafone Idea Q2 FY23 concall), even as its ARPU rose to ₹131 (Vodafone Idea Q2 FY23 concall). Higher price per user, fewer users. (Illustration of how to read the numbers, not a view on the stock.)

That single exchange contains the entire discipline of analysing an Indian telecom. This is a business where the income statement is almost a distraction, because the two forces that decide the outcome — how much each user pays you, and whether you have the capital to keep them on a working network — sit in the operating metrics and the balance sheet, not in the profit line. A telco can report rising revenue per user and be dying. It can post a loss for years and be compounding. The craft is learning which numbers carry the signal.

A note on the boundary first: you will not model a national network from the outside. What you can do is read the direction of ARPU, subscribers, capital intensity and the spectrum-and-AGR overhang, and check whether management's account survives the Q&A.

Why is a telco a capital-eating utility, not a tech stock?

Strip an Indian telecom down and it is a utility with a tollbooth on data. It builds an enormously expensive network — towers, fibre, spectrum bought from the government for tens of thousands of crores — and then collects a small monthly rent from a vast number of people. The economics are brutal and beautiful at once: once the network is built, an extra subscriber costs almost nothing to serve, so incremental margins are huge; but the network is never finished, because every few years a new generation (3G, 4G, now 5G) demands another round of spectrum auctions and capex. Beyond the two giants, the listed field thins quickly to names like Bharti Hexacom, Tata Communications and the state-owned MTNL.

That framing fixes the beginner's error. Telecom is not a growth story driven by adding customers — India is already saturated at over a billion connections. The game is no longer more users; it is more revenue per user on a network you have already paid for. Which is why the single most important number in the sector is the one most casual investors never look at: ARPU.

The metrics that matter

ARPU — the one lever that moves everything

Average revenue per user is exactly what it says: total mobile service revenue divided by the number of subscribers, expressed per month. In a saturated market with fixed costs, ARPU is the lever — almost every rupee of ARPU increase falls to EBITDA, because the network serving that user is already built and paid for.

You will find ARPU in the investor presentation and the concall, not the statutory income statement — companies report it as an operating metric, so it lives in the PPT and the management commentary. Watch the gap between players. In Q3 FY26, Bharti Airtel's India mobile ARPU was ₹259 (Inve data, Q3 FY26), up from ₹245 a year earlier (Inve data, Q3 FY25). Vodafone Idea's ARPU in Q2 FY26 was ₹180 (Inve data, Q2 FY26). Same customers, same country, same tariffs — a ₹79 gap. That gap is the whole story: Bharti's subscribers skew premium and postpaid (postpaid was 57% of its net adds in Q1 FY26 — Inve data, Q1 FY26), so each one pays more. There is no universal "good" ARPU number; the discipline is to read the trend and the gap to the better operator, because ARPU rising while subscribers fall is a very different animal from ARPU rising while subscribers grow.

Subscribers and net adds — quality over count

A net add is the change in subscriber base in a quarter — gross additions minus those who left. But the headline number lies if you don't split it. Bharti added 4.4 million mobile subscribers in Q3 FY26 (Bharti Airtel Q3 FY26 concall) and is gaining premium users; Vodafone Idea lost 0.5 million subscribers in Q1 FY26 (Inve data, Q1 FY26) and 6 million across one FY23 quarter (Vodafone Idea Q2 FY23 concall). Find net adds in the concall and PPT. What "good" looks like: positive net adds skewed toward postpaid and data users, because those are the high-ARPU, low-churn customers. A telco shedding low-ARPU 2G users while adding postpaid is upgrading its base; one bleeding subscribers across the board is melting.

Data usage and the 4G/5G mix — the demand under the price

Data consumption per user, and the share of subscribers on 4G/5G, tell you whether the franchise is healthy underneath the price. A user on 4G or 5G uses far more data, is stickier, and pays more — so the mix shift is the engine of organic ARPU growth even before tariff hikes. Bharti reported 181 million 5G customers and 74% 5G population coverage by Q3 FY26 (Inve data, Q3 FY26). Vodafone Idea's 4G subscriber base was 127.8 million in Q2 FY26 (Inve data, Q2 FY26) with data traffic up 21.4% year on year (Inve data, Q2 FY26) — demand is there; the constraint is network capacity, not appetite. These numbers live in the PPT, never the income statement.

Capex and spectrum — the cost of staying in the game

This is where the cash goes. Capex builds and densifies the network; spectrum is bought at auction and sits on the balance sheet as a deferred liability. The metric that matters is capex intensity — capex as a share of revenue — and its trend. Bharti guided its overall capex lower than FY25 (Bharti Airtel Q4 FY25 concall, marked "diluted" in Inve's Promise Tracker) — a company at the top of its 5G build, harvesting cash. Vodafone Idea, by contrast, guided ₹7,500–8,000 crore of capex for FY26 (Vodafone Idea Q2 FY26 concall) and a three-year plan of ₹50,000–55,000 crore "once we tie up the bank debt" (Vodafone Idea Q1 FY26 concall) — capex it cannot fully fund, which Inve's Promise Tracker marks "at_risk." The tell: a telco increasing capex intensity is either investing for growth or running to stand still — the balance sheet tells you which.

Churn — the leak the ARPU headline hides

Churn is the rate at which subscribers leave. It is the most dangerous metric in telecom because it compounds in the wrong direction. The classic trap: a telco raises tariffs, ARPU jumps, the headline looks great — and the price-sensitive users walk. Vodafone Idea lived this. After its November 2021 tariff hike, ARPU rose 16% on an equal-day basis (Vodafone Idea Q4 FY22 concall), but management noted the subscriber base "declined... because of the tariff interventions especially on the entry-level plans where we have seen maximum churn" (Vodafone Idea Q4 FY22 concall). The deeper, fatal version is the funding-driven churn quoted at the top: when you can't build the network, coverage gaps push users out, which cuts revenue, which starves capex further. Find churn in the concall and PPT; what "good" looks like is low and stable churn alongside ARPU growth — that combination means pricing power, not extraction.

AGR liabilities — the regulatory anvil

This is the metric with no equivalent in any other sector, and it has destroyed more telecom equity in India than any operational mistake. Adjusted Gross Revenue (AGR) is the base on which telcos owe the government licence fees and spectrum charges. After a 2019 Supreme Court ruling vastly expanded the definition, operators were hit with enormous back-dues. For Vodafone Idea these liabilities are existential. As of June 2022, its gross debt was ₹1,990.8 billion, comprising deferred spectrum obligations of ₹1,166 billion and an AGR liability of ₹672.7 billion (Vodafone Idea Q1 FY23 concall) — and the AGR figure kept climbing, to ₹685.9 billion a quarter later (Vodafone Idea Q2 FY23 concall) as interest accrued. You will find the AGR and deferred-spectrum liability broken out in the concall, the PPT, and the balance-sheet notes of the annual report — read all three. What "bad" looks like: a liability that grows on its own through accruing interest, on a balance sheet that cannot generate the cash to service it.

How do you value a telecom company?

Here is the homely way to see it. A telco is a leaky bucket carried uphill. The water in the bucket is your subscriber base; the size of the leak is churn; the rate you carry it up is capex; and there's a man at the top of the hill — the government — holding a bill (AGR and spectrum) you have to pay before you keep a drop. The price of the water is ARPU. Two operators can have identically full buckets today; the one with the smaller leak, the lighter bill, and the strength to keep climbing is worth multiples of the other.

That is why the right multiple for a telecom is EV/EBITDA, not the price-to-earnings ratio. Earnings are routinely negative or distorted by one-off spectrum charges, depreciation on a vast asset base, and finance costs on AGR debt — Vodafone Idea has posted losses every quarter shown above, so a P/E is meaningless (Inve data). Enterprise value over EBITDA captures the operating cash engine and the debt, which is the entire point in a capital-heavy, debt-laden sector. The complementary lens is EV per subscriber — what the market is paying for each customer on the network — which lets you compare a profitable operator against a loss-making one on the same footing.

Crucially, EBITDA must be read against the AGR-and-spectrum liability. A telco at a low EV/EBITDA may look cheap until you notice nearly all that EBITDA is spoken for by debt service and government dues. Bharti's net-debt-to-EBITDA was a comfortable 1.02 (Bharti Airtel Q3 FY26 concall) — it can fund 5G and service its dues and pay dividends. That ratio, not the P/E, is where the survival question is answered.

A tower company like Indus Towers is not valued like an operator: its economics turn on the tenancy ratio — tenants per tower, 1.62 in Q3 FY26 (Inve data, Q3 FY26) — because each extra tenant on the same tower is almost pure margin, which is why Indus runs a ~55% EBITDA margin (Inve data, Q4 FY26). Same principle, different asset: utilisation.

A worked case: said versus did

Take Vodafone Idea through its own words, because the gap between what management guided and what the cash allowed is the clearest teaching case in the sector (illustration, not a view on the stock).

In the FY22 calls, management was explicit that the fix was tariff hikes and fresh funding. ARPU did exactly what they said it would — it climbed quarter after quarter, ₹109 pre-hike to ₹124 to ₹128 to ₹131 across FY22–FY23 (Vodafone Idea Q4 FY22 and Q2 FY23 concalls), "5 consecutive quarters" of growth, in management's own count (Vodafone Idea Q2 FY23 concall). On the single metric they controlled — price — they delivered.

But the metric they could not control told the real story. Over the same stretch, the subscriber base fell from 243.8 million to 234.4 million (Vodafone Idea Q4 FY22 and Q2 FY23 concalls), and the gross debt — deferred spectrum plus AGR — rose from ₹1,978.8 billion to nearly ₹2 trillion (Vodafone Idea Q4 FY22 and Q1 FY23 concalls) as interest accrued faster than the company could pay. The MD's own diagnosis, quoted at the top, closed the loop: churn was rising "because the funding [was] not being available today" (Vodafone Idea Q2 FY23 concall). The capex needed to plug the leak depended on funding that depended on confidence that depended on… not leaking. A telco can win on the lever it controls and still lose, because the bill at the top of the hill keeps growing and the bucket keeps draining.

Inve's Promise Tracker shows the same pattern in the guidance record: the FY26 three-year capex plan sits "at_risk" and the H1 FY26 capex commitment was marked "missed" (Inve data) — the build kept slipping against the funding. This is a read on how management has communicated through a funding crunch, not a lifetime verdict.

Red flags specific to telecom

• ARPU rising while subscribers fall. The headline-flattering trap. Sustainable ARPU growth comes from mix shift and pricing power; extractive ARPU growth comes from squeezing a shrinking base. The subscriber line tells you which.
• An AGR or spectrum liability that grows on its own. When deferred dues climb through accruing interest faster than EBITDA grows, the balance sheet is quietly compounding against the shareholder — even in a quarter with "good" operating numbers.
• Capex guidance hedged on funding it doesn't have. "We'll spend ₹50,000 crore once we tie up the debt" is not a plan; it's a hope. Capital starvation in a network business is terminal because coverage gaps drive the churn that cuts the cash that funds the capex.
• Net-debt-to-EBITDA climbing while the sector's leader's is falling. In a duopoly-ish market, the gap between the strong and weak balance sheet widens through every capex cycle.
• Net adds that aren't split by postpaid/prepaid or 4G/2G. Management showing only the aggregate is often hiding a deteriorating mix underneath a flat headline.

Frequently asked questions

The discipline comes down to refusing to be impressed by the price-per-user headline. The bucket, not the tollbooth, is the business — and it speaks through subscribers, churn, capex intensity, and the size of the government's bill. So invert the question you bring to a telco's results. Don't ask "is ARPU rising?" Ask: if this company were slowly losing the ability to fund its own network, what would the numbers look like — and does this balance sheet rule that out? Rising ARPU on a falling subscriber base, with capex guided on funding it hasn't secured, does not rule it out; it is the pattern itself. Compare that record against the concall guidance and the operating KPI trends, and the same instinct you'd bring to reading an NBFC's balance sheet carries over: the headline is the easiest number to manufacture.

Where this lens can be wrong. The strongest case against everything above is that the weak operator is a survival option, not a melting ice cube — that with one government relief package (converting AGR dues to equity, a moratorium, a tariff floor) and one funding round, the same churning subscriber base inflects, and a stock left for dead re-rates violently because three-player markets are more rational than four. That case has real merit: India's government has repeatedly signalled it wants three private operators, not two, and has already offered moratoriums and equity conversion. Reading the operating metrics would have kept you cautious on Vodafone Idea through the decline — correctly — but it would not have told you the timing of a policy rescue, which is a political event no transcript can forecast. So the honest claim is narrow: the metrics tell you which operator is healthier and which is dependent on outside help. They cannot tell you whether, or when, that help arrives.

And the owner's question, the one to sit with before buying a single share of any telco: what must I believe about the next spectrum cycle — not this quarter's ARPU — for this company to still be funding its own network, servicing its dues, and keeping its best customers on the other side of it? If the honest answer leans on a government bailout rather than the balance sheet's own cash, you have read the tollbooth, not the bucket.

Related reading

• How to Analyse a Bank Stock (NIM, CASA, GNPA, ROA)
• How to Analyse a Building Products Company (Pipes, Tiles)
• How to Analyse a Cables and Wires Stock