ARPU: who counts as a user, and why the debate is about the denominator

ARPU is a service's revenue for a period divided by the average number of users over the same period. The formula is trivial, so the analyst's real work lies elsewhere: in setting two boundaries. The first is which part of revenue goes into the numerator. The second is who exactly the company is prepared to call a user. Issuers make these choices in their own way, there is no single disclosure standard, and the ARPU of two companies cannot be compared head-to-head until you have checked the definitions in each company's own releases.

The numerator: not all revenue indiscriminately

ARPU is almost never calculated from a group's consolidated revenue. Analysts take the revenue of a specific segment, such as mobile services, a subscription service or advertising inventory, and only the portion of it that users generate. One-off equipment sales, interconnect revenue from other operators' traffic and intersegment transactions are usually excluded, because they do not scale with the subscriber base. If a company does not disclose this, comparing its ARPU with a peer's becomes meaningless: you are dividing different numerators by different denominators and getting a difference that does not exist in the business.

VAT and other indirect taxes are a separate issue. Revenue under IFRS is reported net of them, but operating press releases sometimes give "subscriber billings" with the tax included. The difference is small by nature, but large enough to blur the quarter-on-quarter trend.

The denominator: who is a user

This is the main source of discrepancies. The definitions that actually appear in disclosures include:

• SIM cards or connections. Every active line is counted. A person with a work SIM and a personal SIM already counts as several users, and ARPU falls while revenue stays the same.
• Unique subscribers. Lines are collapsed down to the individual. ARPU is higher, but the deduplication methodology is internal and cannot be verified.
• Active over the period (MAU, DAU). The company sets the activity threshold itself: logging into the app, making a transaction, viewing content. Moving the threshold changes the denominator and nothing else.
• Paying users. This is no longer ARPU but ARPPU, a metric of wallet depth rather than of how well the base is monetised.
• Average for the period or end of period. When the base is growing fast, an end-of-quarter denominator understates ARPU; the average gives a fairer picture.

ARPU rises, revenue falls — and there is no contradiction

ARPU multiplied by the size of the base gives segment revenue, so the metric only makes sense alongside the trend in the base. The churn of low-revenue users mechanically lifts ARPU even as revenue shrinks. The opposite picture is aggressive expansion into cheap tariffs: the base grows, revenue grows, ARPU declines. Neither scenario is good or bad in itself; what is bad is an analysis that looks at only one of the components of growth.

For corporate finance, this yields a practical rule: ARPU is not a target but an element of decomposition. In a revenue model it sits alongside net subscriber additions, churn and the tariff mix, and forecasting it in isolation from them is pointless.

What ARPU does not show

This is a top-line metric. It says nothing about the cost of acquiring a user, the user's lifetime or the cost of serving them. A segment with high ARPU can be loss-making if acquisition is expensive and retention is short. That is why, in corporate analysis, ARPU is placed next to measures of cost efficiency — for example, the cost-to-income ratio discussed in the article on CIR and its contested denominator — and next to segment margins.

How to read ARPU in financial reports

Step 1. Find the definition of the base and of the revenue composition in the release — usually in the footnotes to the operating metrics. Quarterly and annual disclosures by Russian issuers are collected in the financial reports section.

Step 2. Check whether the methodology has changed. A footnote about a restatement matters more than the figure itself.

Step 3. Rebuild the product: ARPU multiplied by the base should roughly match segment revenue. If it does not, the numerator is broader or narrower than you assumed.

Step 4. Compare only within the same definition: a company against itself over time, and competitors only once you are sure their bases are counted the same way.

Step 5. Move on to valuation. The issuer pages for companies with telecoms and internet revenue — MTSS, RTKM, OZON — and a multiple such as 9,14 show how the market has already priced the monetisation of the base. The full list of securities is in the Russian stocks section, and definitions of related metrics are in the glossary.

One separate caveat: there is no single database of ARPU values for Russian issuers, because the metric is not part of mandatory IFRS disclosure. Any series you see in a comparison table has been assembled by hand from press releases — and inherits all the differences in definitions described above.