Subscription revenue = installed base × ARPU, and the base stays put unless the customer acts to leave. Staying is passive and leaving takes effort. That flipped default is worth an enormous multiple premium when churn is low, because year-N revenue arrives without year-N selling cost. The P&L has a high gross margin, since serving one more subscriber is cheap. It also carries heavy upfront spend to win customers and to build what they subscribe to (content, software, a fund-accounting platform). Profitability ends up a function of age and churn. Young subscription businesses look terrible because they're all CAC on a small base. Old low-churn ones look magical. Cash usually leads revenue. Customers pay for the year and the company recognises it monthly, so the unearned balance sits as deferred revenue / contract liabilities. Read that line as a leading indicator.
Subscription business model examples
Netflix (US). At an illustrative ~2024 scale, ~302m members × ~$12/month blended gives revenue of ~$39bn. Operating margin is ~27%, up from ~13% in 2019, so hold that thought until operating leverage comes up. Cash content spend runs ~$17bn/yr, amortised at ~$15bn through the P&L, and FCF is ~$7bn. In one line, content is a fixed cost and subscribers are the variable revenue. Every extra member streams the same library at ~zero marginal cost, so margin expands mechanically as the base grows. That holds provided churn stays low and content spend need not scale with members. The bear case is always one of those two provisos.
Advertisement
The Indian annuity-style family. India's cleanest recurring businesses are toll-keepers, paid in basis points on pools of assets that grow by themselves. CAMS is registrar to ~68% of mutual-fund AUM, with revenue ≈ AUM × ~3bps blended. That comes to ~₹1,400 cr of revenue at a ~33% PAT margin (illustrative ~FY25). HDFC AMC and its peers charge fees in bps on AUM. CDSL levies annual issuer charges per folio. Depositories and exchanges generally work the same way. The subscriber never "renews". The SIP auto-debits, the AUM compounds with the market, and the fee meter keeps running. Churn is structurally near zero because switching an RTA or depository is an institutional migration project. So when an Indian analyst note calls something "annuity-like", this is the pattern being claimed. Your job is to test the two walls holding it up: is the pool sticky, and is the fee rate defensible? Fee-rate regulation, such as SEBI's TER cuts, is this family's version of churn.
The B2B maintenance annuity. Take Otis (US, illustrative ~2024). New-equipment revenue is ~$5.6bn at ~6 to 7% operating margin. Service revenue is ~$8.7bn at ~24 to 25%. Otis sells the elevator near cost to install a 30-year maintenance subscription, and service is ~85% of segment profit. India runs on the same physics wherever an installed base needs mandatory care: lifts, gensets, medical devices, even auto service networks.
What should you check in the filings?
What to check in filings: net adds and churn first, or its proxies where churn isn't given: renewal rate, persistency, gross vs net retention. Then check whether the ARPU trajectory is price-led or mix-led. Compare deferred revenue growth with revenue growth, because decelerating bookings show up there first. Read the RPO (remaining performance obligations), which is contracted future revenue disclosed under Ind AS 115/ASC 606. Finally, set S&M as a % of revenue against base growth. Are they buying the growth? The unit economics module puts a number on it.