Subscribe
How Apps WinFacts as of

Why Your Freemium App Isn’t Making Money (And How To Fix It)

The short answer

Hard paywall vs freemium: which earns more per download?

In RevenueCat's 2026 data, the median hard paywall app earned $3.09 per download by day 60, against 38 cents for the median freemium app: about 8x. If your app shows its value in the first session, ask for money in the first session. Then give value before you ask for an account, personalise before the offer, price at the going rate and protect the first week.

Key facts

Chapters

  1. 0:00Two identical apps
  2. 0:31Meet the twins
  3. 0:49Decision 1: The first minute
  4. 1:38Decision 2: Three questions
  5. 2:21Decision 3: The offer
  6. 3:55Decision 4: The price
  7. 4:53Decision 5: The first week
  8. 5:50The scoreboard
  9. 6:23What happens when users say no?

Transcript

Edited for reading

Grey and Amber are fictional habit tracker apps. Each one gets the median result from real 2026 industry data, listed under Sources.

Two identical appsWatch from 0:00

These two apps are identical. One made $3,800 in its first 60 days. The other, almost $31,000. 8x more.

Same idea. Same features. Same launch day. Same 10,000 downloads. We ran both through real benchmark data, and the features were never the difference.

The difference is five decisions, and most teams make them without noticing. One of them lines up with an 8x gap across thousands of real apps. Let's take them one at a time.

Meet the twinsWatch from 0:31

Quick note: Grey and Amber don't exist. We built them, and gave each one the median result for apps like it. The data comes mostly from RevenueCat and Adapty, and every source is listed below.

Both are habit trackers. We'll follow their 10,000 users through five moments, starting the second someone opens the app.

Decision 1: The first minuteWatch from 0:49

Grey opens with a form. Email. Password. Confirm password. It feels responsible. You'd want an account before you save anything, right?

Amber opens with one question: what habit do you want to build? You pick "read more books", and seconds later you're setting your first goal. Amber asks for an account later, once there's something worth saving.

That order matters. For the median app, only 26% of new users come back the next day. By day 30, it's 7%. Every screen between a new user and the first useful moment is a chance to lose them.

Onboarding tool makers report 10 to 30% more users getting started when sign-up comes later. It's a vendor number, not a study, but it's cheap to test.

Decision one: give before you ask. Grey asked first. Amber gave first.

Decision 2: Three questionsWatch from 1:38

Amber's next move looks small. It asks three quick questions. What's the goal? When in your day does it fit? What stopped you last time? Then it shows a plan: your plan, at your time, starting tonight.

Grey skips all that and drops you on an empty screen. No habits yet. Tap plus to add one.

In a study of 1,460 onboarding flows, only 23% of apps asked personalising questions at all. Most apps skip this step entirely.

Those questions do two jobs. The user feels the app will work for them before they've even used it. And the app learns exactly what to say at the next moment, the moment where the money is made.

Decision two: make it theirs before you make the offer.

Decision 3: The offerWatch from 2:21

Now the offer. When does an app ask for money? Almost 9 in 10 free trials start on the day of install. Partly that's because it's when most apps ask. But it means an offer hidden in settings is one almost nobody takes.

Amber shows its offer right after the plan, and there's no free version behind it. To start the plan, you start a free trial. That's called a hard paywall. Grey is free, with a premium button tucked away in settings.

By day 35, the median hard paywall app has turned about 1 download in 10 into a paying customer. The median freemium app, about 1 in 50. For our twins, that's 210 paying customers for Grey, and 1,070 for Amber.

In revenue, the median hard paywall app earned $3.09 per download by day 60. The median freemium app earned 38 cents. Multiply by 10,000 downloads, and you get the numbers from the start of this video.

The obvious worry is that a hard paywall gets you worse customers. The data says no. After one year, hard paywall apps kept 27% of their yearly subscribers. Freemium apps kept 28%.

To be fair, this compares different apps. It isn't an experiment. Freemium apps also keep converting for longer, so a 60-day window flatters the hard paywall. And apps that need a network of users, earn from ads, or take weeks to show their value can be right to stay free.

So the lesson is narrower: if your value shows up in the first session, ask in the first session.

Decision three: ask for the money up front. That's the choice that lines up with the 8x gap. And Amber is still leaking money in two places.

Decision 4: The priceWatch from 3:55

Now both teams face the same question: what to charge. Grey goes cheap: $19.99 a year, well under the going rate, so nobody's put off. Amber charges about the market median, $34.99, and doesn't apologise.

You'd expect the cheap app to get more people through the door. In RevenueCat's data, it's the other way round. High-priced apps turned downloads into free trials at about twice the rate of low-priced ones.

And after a year, a paying customer of a high-priced app was worth about $62. For a low-priced app, about $11. Most of that gap is simply the price: each customer pays more. In this data, the cheap price didn't bring in more trials. It just made each customer worth less.

Part of the gap is also that better apps tend to charge more. So this isn't advice to double your price tomorrow. It's advice to treat price as a decision you test, not a number you picked because it felt safe.

Decision four: charge what it's worth, then test it.

Decision 5: The first weekWatch from 4:53

Last decision. The customer has started a trial. Most teams celebrate and move on.

But of all cancellations on 3-day trials, 55% happen on day zero, the same day the trial started. Yearly plans aren't safe either: 35% of all annual cancellations land in the first month. Many people pay, then switch off auto-renew straight away, just in case.

Grey goes quiet after the sale. Amber treats the first week as part of the product, starting on day zero: a first win before the user closes the app, and a reminder before the trial ends, showing what they've built.

And some customers don't choose to leave at all. On Google Play, 31% of subscription cancellations are failed payments. On the App Store, 14%. Often it's just an expired card. So Amber turns on grace periods and billing retries, and wins back some of the customers who never meant to leave.

Decision five: the first week turns a sale into a customer.

The scoreboardWatch from 5:50

So here are the twins side by side. Grey asked first, showed an empty screen, hid its offer, priced cheap and went quiet. Amber did the opposite, five times over.

We gave each twin the median result for its paywall type: $3,800 against almost $31,000, from the same 10,000 downloads.

The other four decisions don't come with a clean number, because the data can't separate them out. But each one plugs a leak you saw today.

Apps don't win on features. A feature can be copied in a week. They win on decisions their rivals never realised they were making.

What happens when users say no?Watch from 6:23

The paywall screen itself, and what the best apps do about the people who say no, is the subject of Why Your Paywall Design Is Losing Sales (And How To Fix It).

Sources

  1. RevenueCat, State of Subscription Apps 2026 (March 2026)
  2. Adapty, State of In-App Subscriptions 2026 (March 2026)
  3. Adjust, What makes a good retention rate
  4. Daniliants, I studied 1,460 onboarding flows (undefined 2025)
  5. UXCam, onboarding data from Appcues

Caveat: The paywall comparison is between different apps, not an experiment, and a 60-day window flatters the hard paywall. RevenueCat's chart puts the first month's share of annual cancellations at 34%; its report text says 35%, the figure used in the video.

Watch next