Product ownership · Monetization

Paywall experiments that paid off

Out of 100+ A/B tests across 20+ subscription apps, two that show how I work on the moment people decide to pay: how the paywall shows the product, and how it charges for it.

Method
A/B tests in Adapty
Apps
Music and utility
Outcome
×2.2 revenue per user

Overview

One question, two levers

The paywall is where a free user decides whether the product is worth paying for. These experiments asked what makes that decision easier: seeing the value more clearly, or having more ways to pay for it.

Experiment 1 · Presentation

A karaoke app with autotune

A new paywall structure that shows what the app does for your voice.

Experiment 2 · Plans

A phone cleaning app

Monthly and yearly plans added next to the only plan it had: weekly.

Experiment 1 · Presentation

A paywall that shows the product

A karaoke app for singing along to backing tracks with autotune, sold on weekly, monthly and yearly plans. Its long paywall led with an illustration of headphones and listed everything at once.

Control: headphones illustration, headline and a plain feature list
Control: headphones and a long feature list
Variant: a studio microphone, the headline A Pro Studio in Your Pocket and a step-by-step list of features
Variant: the microphone and a guided structure

Hypothesis

If users see a high-quality paywall that shows what the product does, more of them will decide to pay, because a clear reason makes the decision easier.

What changed

The image moved from listening to singing: a studio microphone instead of headphones. Each feature was rewritten around what it does for your voice, in a guided order, then social proof and the plans. Same plans and prices.

Test design

All users, 14 days, 50/50. Primary metric: trial conversion. Guardrails: purchase conversion and revenue.

+10.9%

revenue per 1,000 users.

+11.6%

purchase conversion.

≈ 0%

trial conversion, 10.5% against 10.4%.

67%

probability that the variant is the better one.

Reading the result

Trials didn't move; buyers did. The results were read months after the test, with renewals and refunds already in. 67% is a lean, not proof, but every metric pointed the same way and none got worse over time, so the variant became the default.

What came next

The US moved to its own track: markets read a paywall differently, so the paywall was tested again there on its own, followed by small step-by-step price increases on every plan.

Experiment 2 · Plans

More ways to say yes

A phone cleaning app for clearing files off your phone. Because of what it does, it only offered one plan: weekly, with a three-day free trial.

Control: the plan section with a single weekly plan and a three-day free trial
Control: one weekly plan
Variant: weekly, monthly and yearly plans, with the weekly plan selected
Variant: weekly, monthly and yearly

Hypothesis

With more plans, people would either pick a more flexible option or go straight to the recommended one. Either way, more of them would buy.

What changed

Only the plans. Monthly and yearly were added next to the weekly one, which kept the only free trial, and the paywall design stayed exactly the same. Changing both at once would have made it impossible to tell which one caused the result.

Test design

All users, 6 days, 50/50. Primary metric: revenue per user. Guardrails: purchase conversion and refund rate.

×2.2

revenue per 1,000 users.

+39%

purchase conversion.

−88%

refund rate: 2% against 18% in the control.

98%

probability that the variant is the better one.

Reading the result

Every purchase in the test was still weekly. The monthly and yearly plans worked as anchors: next to them, the weekly plan looked like the easy choice. The results were read months after the test, with renewals and refunds already in, so the guardrails are part of the result, not a follow-up.

What came next

The new plans shipped. Then the same structure was tested across the other apps in the portfolio, to learn where it works and where it doesn't.

Learn

What they taught me

Presentation decides

How a paywall shows the product changes how quickly people decide to buy, even with the same plans and prices.

Options change behavior

More plans changed how people bought, even though every buyer still chose weekly. Choice is psychological, not just practical.

Read past the primary metric

Experiment 1 won on purchases, not trials. Guardrails are where the real result often shows up.