पाठशाला Pathshala · उत्पाद Utpād, The product · Lesson 11 · Build
Freemium, free trial or neither
The conversion model is decided by three numbers: what it costs to acquire a customer, how long a user takes to reach a real result, and how much value the product shows in its first week.
Pathshala, The Founder Library · 11 October 2026 · 7 min read
When Dropbox filed to go public in 2018 it reported that, as of the end of 2017, it served over 500 million registered users but only 11 million paying users, and it warned investors that a majority of registered users may never pay. That is what freemium looks like when it works. Most companies that copy the model have the free users and not the eleven million.
The choice between a free plan, a free trial and neither is usually made by imitation: the competitor has a free plan, so we have one. This lesson makes it from three numbers instead. It defines the three models, gives the conversion rates that count as good for each, sets out the arithmetic of carrying free users, walks the decision as a tree, and covers the Indian rules on card-on-file trials that change how a trial ends.
Three models, defined
Freemium is a free plan with no end date, limited in some way, alongside paid plans. The free plan is a distribution channel: users arrive, some succeed, and success pushes them into the limit. A free trial gives full or near-full access for a fixed period, usually seven to thirty days, after which the user pays or loses access. It comes in two forms: opt-in, where no card is taken and the user chooses to pay at the end, and opt-out, where a card or mandate is taken at the start and charged unless the user cancels. Neither means the customer pays before using the product at full scale: a paid pilot, an onboarding fee, a first month up front with a refund promise, or a demo followed by a contract.
The models are not stages a company passes through in order. A product can run a free plan for individuals and a sales-assisted trial for teams, and many do. But each audience needs one model, chosen on purpose.
Two hybrids are worth knowing by name. A reverse trial gives every new account the paid plan for a fixed period and then drops it to the free plan rather than locking it out, so the user has felt the paid features before deciding and still has a reason to stay. A usage-capped plan charges nothing until a threshold of documents, messages or transactions and then charges by use, which suits products whose value scales with volume. Both are combinations of the three models, not escapes from the choice, and the same three numbers decide whether they work.
What good conversion looks like
Kyle Poyar and Lenny Rachitsky’s 2023 survey of more than a thousand products, mostly B2B software, measured the share of new accounts that start paying within their first six months. For self-serve freemium, good was 3 to 5 per cent and great 6 to 8. Freemium with a sales team helping did better, 5 to 7 per cent good and 10 to 15 great. Free trials converted at 8 to 12 per cent for good and 15 to 25 for great, and trial companies were far more likely to have sales contact most of their sign-ups. David Skok’s SaaS metrics guide puts the ideal trial conversion at around 15 to 20 per cent.
The numbers are not comparable across models, and founders often compare them anyway. A freemium product converting 4 per cent of a large free base may make more money than a trial converting 15 per cent of a small one. What matters is revenue per sign-up, the cost of the sign-ups who never pay, and how fast the paying ones pay back what it cost to acquire them.
The three numbers that decide it
Customer acquisition cost. Skok’s guidelines are that the months to recover CAC should not run beyond twelve, with many of the best SaaS businesses recovering it in five to seven, and that lifetime value should be more than three times CAC. Freemium can lower blended CAC when free users arrive cheaply and bring others; it raises it when every free user is bought through ads and most never pay. Work through [the CAC and payback lesson](/library/cac-ltv-and-payback-the-three-numbers) before choosing.
Activation time: how long a new user takes to reach the first real result, as defined in [the activation lesson](/library/activation-first-session-that-decides). If it happens in the first session, both freemium and a short trial can work. If it takes days, a trial must be long enough to contain it, and a seven-day trial for a product that takes ten days to show value converts only the unusually fast. If it takes weeks or needs setup, free access mostly produces accounts that never started.
Value shown in a week. Some products deliver almost everything they ever will in the first week: a design tool, a converter, a calculator. Others build value over months as history accumulates: accounting, analytics, a customer database. The first kind suits a trial, because the user can judge the whole product in the time allowed. The second suits freemium with a limit on the thing that accumulates, because the upgrade arrives exactly when leaving has become expensive.
The arithmetic of free users
A Kochi company sells scheduling software to tuition centres at ₹1,200 a month. Its marketing brings 10,000 sign-ups a month to a free plan. If 4 per cent pay within six months, each month’s cohort adds 400 paying centres, about ₹4.8 lakh of monthly revenue. But the free plan sends WhatsApp reminders to students, and each active free centre costs about ₹20 a month in messages and servers. If the active free base settles at 25,000 centres, the plan costs ₹5 lakh a month, more than a month’s cohort brings in.
The same company with a fourteen-day trial and no free plan might see only 4,000 trial starts from the same traffic, because a deadline puts people off. At 12 per cent conversion that is 480 paying centres, ₹5.8 lakh of new monthly revenue, with no free base to carry. The figures are illustrative, but the shape is common: freemium wins only when free users cost close to nothing or bring in other users who pay. Message-heavy products in India, where every reminder has a per-message cost, often fail the first test.
A free plan is a channel with a cost per paying customer. If you would not buy that customer through an ad at that price, do not buy them through a free plan.
Trials with a card, in India
Opt-out trials, where a card is taken at sign-up and charged at the end, lean partly on inertia: the charge goes through unless the user acts. In India two sets of rules make that mechanism weaker and its abuse risky. RBI’s e-mandate circular for cards requires additional authentication to register a recurring mandate and a notice to the customer at least 24 hours before each debit, with the amount and the merchant’s name, and lets the customer cancel from that notice. Every first charge is therefore preceded by a reminder. Design for it: send your own message first, saying what the customer achieved in the trial and what the plan costs, rather than leaving the bank to deliver the news.
The Central Consumer Protection Authority’s Guidelines for Prevention and Regulation of Dark Patterns, 2023, name subscription trap and SaaS billing among thirteen specified patterns, and in June 2025 the authority asked e-commerce platforms to audit themselves for them within three months. A trial that is hard to cancel, that hides the auto-renewal or that charges without clear disclosure is now a compliance problem as well as a reputational one. Make cancelling as easy as signing up, state the price and the date of the first charge on the sign-up screen, and keep both rules in view when you read your conversion rate. Both were checked in October 2026.
Changing the model later
Moving from a trial to freemium is easy: you are giving something away. Moving from freemium to a trial is hard, because you are taking it back from people who were promised it, and the lesson on [pricing experiments](/library/pricing-experiments-without-burning-customers) applies in full: keep existing free users on their plan for a stated period, give notice and offer a bridge. The practical advice follows. If you are unsure, start with a trial or a narrow free plan and widen it later; a generous free plan is very hard to narrow.
Deciding this month
This week, measure three numbers from your own data: the median time from sign-up to the first real result, the monthly cost of serving one free or trial user, and blended CAC against monthly revenue per paying customer. Walk the tree with them. Write the chosen model, the audience it applies to and the conversion rate you expect in one paragraph. Run it for two full cohorts, at least two months, and compare revenue per sign-up and the cost of non-paying users against the paragraph. Then review once a quarter, and walk the tree again whenever activation time or CAC changes by more than a third.
Nothing here is legal, tax or investment advice. The worked company is illustrative; the benchmarks and rules are from the sources below.
Sources
- Dropbox, Inc., Form S-1 registration statement, February 2018
- Kyle Poyar and Lenny Rachitsky, What is a good free-to-paid conversion rate, Lenny’s Newsletter, August 2023
- David Skok, SaaS Metrics 2.0, For Entrepreneurs
- PIB, CCPA advisory to e-commerce platforms to self-audit for dark patterns, June 2025 — Names the thirteen patterns of the 2023 guidelines. Checked October 2026.
- RBI, Processing of e-mandate on cards for recurring transactions, circular of 21 August 2019