Summary

  • The September-quarter pack prints India mobile revenue of Rs 309,431 Mn, a Mobile Services India customer base of 377,224 thousand and an ARPU of Rs 271; dividing the first by the second and then by three months gives Rs 273.43, which is my arithmetic and Rs 2.43 above the printed quotient.
  • The divisor that would reproduce Rs 271 exactly is about 380.6 Mn, roughly 3.4 Mn above the printed quarter-end count and about 5.4 Mn above a two-point average of printed counts; the India-wide base of 498,685 thousand printed in the same quarter's press release is a different perimeter and produces Rs 206.83, not Rs 271.

Three numbers sit on the same page of Bharti Airtel Limited's September-quarter investor pack, and they belong together. Section 4.2, under the heading Mobile Services India, prints a customer base of 377,224 thousand and an Average Revenue Per User of Rs 271, with 376,508 thousand and Rs 264 in the June column and 364,192 thousand and Rs 256 a year earlier. Section 5.5.3.1 prints the segment's revenue and repeats the quotient in a sentence: "Mobile revenues for the quarter stood at Rs 309,431 Mn, up by 10.1% compared to Rs 281,168 Mn in the corresponding quarter last year." A page later the same section writes it plainly: "Overall ARPU for the quarter stands at Rs 271." One document, one segment, one quarter — Quarterly IR Pack, Q2 FY27, Bharti Airtel Limited.

Everything needed for the division is present except the divisor. So do the division.

Rs 309,431 Mn across 377,224 thousand customers is Rs 820.28 per customer for the quarter. Divided by the three months of the quarter, that is Rs 273.43 per customer per month, against a printed Rs 271. The subtraction is Rs 2.43, or 0.90 per cent. This is my arithmetic, not the company's, and it is the only kind of arithmetic a reader holding the pack can perform.

Run the same operation on the year-earlier quarter and the sign does not change. The September 2025 pack prints Rs 256 for the September 2025 column, with 364,192 thousand mobile customers and Rs 250 and 233 in the two comparative columns — Quarterly IR Pack, Q2 FY26, Bharti Airtel Limited. Divide the Rs 281,168 Mn that the later pack prints as the comparative revenue by 364,192 thousand and the quarter yields Rs 772.03 per customer; divide by three and it is Rs 257.34, against a printed Rs 256. That is Rs 1.34, or 0.52 per cent, and it is again mine.

Two quarters pointing the same way is information. Reverse the division and it sharpens. If Rs 271 is the monthly quotient, the customer count behind it must be Rs 309,431,000,000 divided by 271 times three, which is 380,603 thousand — about 380.6 Mn. The same page prints 377,224 thousand. The implied divisor is therefore about 3.4 Mn above the stock it is printed beside. For the year-earlier quarter the implied divisor is 366,104 thousand against a printed 364,192 thousand, about 1.9 Mn above.

A quarter-end stock is a poor candidate for a period ratio in a growing base, so test the obvious alternative. The March 2026 pack prints 373,243 thousand mobile customers at 31 March 2026 and 368,543 thousand at 31 December 2025, with ARPU of Rs 257, 259 and 245 across the three columns — Quarterly IR Pack, Q4 FY26, Bharti Airtel Limited. Average the March and September stocks and the two-point mean for the September 2026 quarter is 375,233.5 thousand. The implied divisor of 380,603 thousand sits about 5.4 Mn above that too. For the year-earlier quarter, averaging 362,796 thousand and 364,192 thousand gives 363,494 thousand, and the implied divisor of 366,104 thousand sits about 2.6 Mn above it. No average built from the printed dates reproduces the printed quotient.

One test does settle part of the question, and it is dimensional rather than documentary. If Rs 271 were a quarterly amount per customer, the divisor would be Rs 309,431,000,000 divided by 271 — 1,141,812 thousand customers, about 1,141.8 Mn, roughly three times the 377.2 Mn the same pack prints for the same segment. Only a per-month reading survives contact with the printed stock. The line does not say per month; it is labelled Rs, and that is the whole of the label.

The trap next door is the perimeter. The September-quarter press release prints an India customer base of 498,685 thousand at 30 September 2026, 491,890 thousand at 30 June 2026 and 449,738 thousand a year earlier, in a table whose column also carries the mobile improvement quoted in its own text, where ARPU "expanding to Rs 271 in Q2'27 from Rs 256 in Q2'26" — Press Release, Q2 FY27, Bharti Airtel Limited. Divide the same Rs 309,431 Mn by that larger base and the quarter gives Rs 620.50 per customer; divide by three and it is Rs 206.83 a month, 23.7 per cent below Rs 271. The two printed bases are not interchangeable, and only one of them sits in the Mobile Services India table.

Management's own commentary adds a second reason the remaining Rs 2.43 should not be read as an error. On the earnings webinar held on 6 November 2026, the transcript records: "ARPU for the quarter came in at Rs.271, up from 264, benefited by one extra day in the quarter as well." The same passage says India mobile added 0.7 million revenue-earning customers and 2.9 million smartphone data customers during the quarter — Earnings Call Transcript, Q2 FY27, Bharti Airtel Limited. The pack's own table is consistent with the first of those: 377,224 thousand less 376,508 thousand is 716 thousand net additions, which is the 0.7 million. Whether a revenue-earning customer, a smartphone data customer and a mobile customer are one noun is not stated anywhere in the material read, so they are kept apart here.

The preceding quarter supplies a third dated pair, and a third confirmation that the quotient is printed while the divisor is not. The June-quarter press release records ARPU rising "to Rs 264 in Q1'27 from Rs 250 in Q1'26" and prints an India customer base of 491,890 thousand at 30 June 2026 — Press Release, Q1 FY27, Bharti Airtel Limited. The one-page highlights sheet for the same quarter states the same pair in a single line — Quarterly Highlights, Q1 FY27, Bharti Airtel Limited. The rupee amount of India mobile revenue for that quarter was not obtained from a filed document, so this piece does not run the division on June 2026 at all.

What the India documents do not print is the convention. Two other listed companies in the same group do print it, and their wording is worth having precisely because it is not Bharti Airtel Limited's. Airtel Africa plc's 2025 annual report glossary defines ARPU as "Average revenue per user per month. This is derived by dividing total revenue during the relevant period by the average number of customers during the period and dividing the result by the number of months in the relevant period," and defines average customers as monthly averages, each being the mean of opening and closing active customers for that month — Airtel Africa plc, Annual Report and Accounts 2025, glossary.

Airtel Malawi plc's half-year glossary, published through the Malawi Stock Exchange, says ARPU is "Average revenue per user per month, which is derived by dividing total revenue during the relevant period by the average number of customers," and that average customers are "derived by computing the average of the monthly average customers for the relevant period" — Airtel Malawi plc, results for the half year ended 30 June 2026. The two formulations differ in wording, and neither is a statement about India.

Three entities appear in this piece and they are not one actor: Bharti Airtel Limited is the listed issuer whose September-quarter pack carries the ARPU, the customer base and the revenue; Airtel Africa plc is a separate listed company whose glossary is quoted for its wording only; Airtel Malawi plc is a listed subsidiary whose half-year glossary is quoted on the same terms.

The September pack also refers readers to a Glossary at Section 11 for "detailed definitions" of its non-GAAP and operating measures. That glossary body was not obtained from the material read on the date read, so nothing here quotes or paraphrases it.

Six fields cannot be closed from the pages that were read: the glossary's own ARPU definition; a printed average-customer count for Mobile Services India for any quarter; a printed per-month basis beside the India ARPU line; the revenue scope behind the ARPU numerator; a printed exchange rate beside the dollar ARPU; and the rupee amount of India mobile revenue for the June 2026 quarter from a filed document.

The dollar line is its own small illustration of the same habit. The same pack prints the same quarter's ARPU as US$ 2.8 with no rate beside it. Rs 271 against US$ 2.8 implies about 96.8 rupees to the dollar, and because 2.8 is printed to one decimal the implied rate lies roughly between 95.1 and 98.5. That band is mine. No page read for this piece supplies the rate.

So what can a stranger holding the issuer's own pages compose? Three of the six fields, and no more. The quotient as printed, with its currency and its period. The stock as printed, with its date. A division of the printed revenue by the printed stock that lands within a rupee or two of the quotient and never exactly on it. That is enough to show the per-customer figure is a monthly measure, enough to show the divisor is not the count in the same row, and not enough to say which count it is. The company is followed in BTW's directory as an APNIC member and number-resource holder — Bharti Airtel Limited in the BTW directory — and the operating record above is what the entry does not, and should not, pretend to answer.

One number in this piece is checked and reproducible, and it is the one nobody printed: the divisor. Everything else a reader will want to compute from Rs 271 — a monthly value of a customer, a revenue per user per quarter, a comparison of one segment's ARPU with another's — needs a convention the reviewed pages do not state. The pack prints the quotient. The stock in the next column is a photograph of one day. The distance between them, 3.4 million customers in September 2026 and 1.9 million a year before, is the size of the assumption a reader is being asked to make without being told its name.