Give the core product away and charge on transactions

Free core, paid transactions means you charge nothing for the tool people use every day and earn a fee when money moves through it. Wave gave away accounting and invoicing software and later charged for payments, loans and payroll. Square gave away its card reader and took 2.75 percent of each payment. Both grew fast, and neither result was tested against a control.

How it works

The model works when the free tool sits right next to a money event your customer already has. The steps:

1. Give away the tool that creates the money event. Wave offered free cloud accounting and invoicing to small businesses. A newspaper report on Wave's sale says this built its user base quickly and at low cost. See the Wave case.
2. Remove the setup cost too. Square handed out its reader at signup. Early on it also sold the reader in Apple stores for $9.95 and gave new users a $10 rebate.
3. Charge where money moves. Wave first planned to earn from advertising. Around 2014 to 2015 it started charging for bank transfers, loans and payroll built into the product. Square charged a flat 2.75 percent on each payment.
4. Let usage grow revenue for you. Revenue scales with customer volume, not with seats. A profile of Square says it had 1 million users and about $4 million in daily processing by August 2011. It later announced 2 million users and $6 billion a year, over $16 million a day. See the Square case.

When Wave was sold to H&R Block, it had more than 400,000 customers, with tens of thousands paying. Its annual revenue was above $50 million. Wave's about page says it has served over 5.9 million small business owners across the US and Canada since 2010. Square still uses a free entry point. A Square press release describes accepting contactless payments directly from an iPhone with no extra hardware and no added cost to the seller. Square frames this as a way into its wider set of business products.

What it takes

Our estimate: about 80 hours of work, and about 45 days before you can read a signal. Most of the hours go into wiring a payment partner into the step where customers already handle money, setting the fee, and writing the pricing page and onboarding copy. A small team can usually partner for the regulated parts instead of building them. The signal is slow because customers first have to adopt the free tool, then hit a real money event, such as sending an invoice and getting paid. Monthly billing cycles mean you need at least one full cycle plus some margin.

What to measure

The main number is the share of active free accounts that run at least one paid transaction, tracked by signup cohort. The Globe and Mail says Wave's conversion rate is believed to be about two to three times the 2 to 4 percent typical of freemium companies. Our free-to-paid conversion benchmarks can help you place your own figure. A good reading is a share that climbs as cohorts age. A bad reading is a flat share near zero after the first billing cycle.

Next, compare transaction revenue per active account with the cost to serve a free account. If free accounts cost more than paying ones earn, growth makes your losses bigger.

Track signup rate before and after you drop the subscription price, using a fixed window. If signups barely move, the price was not what held people back.

Finally, compare retention of accounts that transact with accounts that never do. Read the curves with our guide on retention curves. Transacting accounts should flatten higher. If both curves look the same, the paid layer is not making the product more valuable to keep.

When it backfires

• The free core never touches money. Wave's first revenue plan was advertising, and fees only arrived years later. If your free tool is far from a payment, you may wait a long time for revenue.
• Most users never pay. Even at its sale, Wave's paying customers were in the tens of thousands out of more than 400,000. Budget for a large free base that you serve at a loss.
• The giveaway has a unit cost. Square gave readers away at signup and offered new users a $10 rebate on the $9.95 Apple store price. The sources give no cost per reader or per new user, so the subsidy cannot be checked.
• Operations grow with volume. Square went from 80 employees to 300 within about a year. Transaction businesses bring support, risk and compliance work that a subscription product does not.
• The free entry point moves. Square now accepts contactless payments on a phone with no extra hardware.

Run it as an experiment

Hypothesis: making the core tool free and adding a paid transaction step will raise total revenue per signup cohort within two billing cycles, compared with the current subscription.

Comparison: we recommend a split by new signup, with half seeing the free core plus fees and half seeing current pricing. If a split is not possible, run a before and after with a fixed eight week window on each side. Use the sample size calculator to check that you have enough signups per arm.

Log signups, activation, first paid transaction date, transaction count and value, fee revenue, subscription revenue and support tickets per account. Put the results into the growth model template to project payback by cohort.

Decision rule: keep the model if revenue per cohort at the end of the window matches or beats control and cost to serve stays below fee revenue. Otherwise roll back. Follow our guide to running a growth experiment for the write-up.

How strong the evidence is

The grade is case with numbers. At least one company documented the tactic and the result was published as numbers. Wave carries most of that weight, but its figures come from a newspaper report made at the time of a sale. The conversion rate is described as believed, not confirmed. There is no baseline, and the revenue is not attributed to the fee layer alone. Square's user and volume figures were announced by the company and reported by a third party. Square's outcome is mixed, and the sources give no unit economics. The sources also disagree on dates: Square's about page places the first reader in 2009, while the profile cited above says the company launched in May 2010. Neither case shows what would have happened with plain subscription pricing.