Why Users Download an App Without Buying It

People download many apps and buy almost none, for reasons a numbered funnel helps locate: the audience, the moment payment is offered, what the free version allows and the price. Here is the real funnel of our Android app, nothing dressed up.

This is the third episode of our series on building a SaaS. The figures come from our PostHog tracking over 28 days, taken in early October 2026, for VoxCut on Android. "People" means installs: one user can count several times. For context, read why a SaaS can have users but no sales.

Over 28 days: 344 installs. 151 people processed a file. 240 saw the payment screen. 26 clicked to buy. No real purchase was recorded. The only purchase event in the data was a licence test we ran ourselves.

Another figure: of 146 people who finished a processing, only 12 came back on a second day.

A funnel does not say why, it says where. Three leads stand out.

• An audience not ready to pay that price. Paywall views come mostly from countries with lower purchasing power: 53 from India, 40 from the United States, then Egypt, Brazil and Bangladesh. The starting price may be too high there.

• A block at the wrong moment. A five-minute duration limit blocked 71 people, only two of whom clicked to buy. A hard block with no alternative discourages more than it converts.

• An occasional need. Few people return the next day: a tool you use once does not call for a purchase.

On the PC version of the same app, someone who exceeds the limit is offered a partial export of the first minutes; on Android, the block was hard. A partial export lets people see the value before being asked to pay. When you block without showing a result, the user has seen nothing that justifies paying. It is the lesson that recurs in every paywall study: timing matters as much as price.

Rather than changing everything, we test one change at a time, to know which one worked.

• A per-country price: in India, the lifetime price was lowered from 1,650 to 249 rupees on 5 October, keeping other countries as controls, with a review planned about four weeks later

• A partial export in place of the hard block, planned for a coming version

• Recording the real file duration at the moment of the block, which was missing for understanding

We will publish the results, good or bad, in a later episode.

• Distinguish installs from real people

• Look at conversion by country, not only overall

• Beware of small numbers: 26 clicks allow no fine conclusion

• Change a single variable at a time

To choose a business model knowingly, read freemium, free trial or subscription.

A download shows curiosity, not purchase intent. The funnel says where people are lost; it is up to us to test why. You can follow VoxCut's progress at voxcutpro.com.

Why do people download without buying? Curiosity, an occasional need, a price unsuited to the country, a block at the wrong moment: a numbered funnel helps see which of these dominates.

What is a normal conversion rate? It varies enormously with the product and how you measure. Compare yourself with your own history rather than averages.

Should you lower the price? Maybe, but test it without changing anything else, and by country if your audience is international.

Is a hard block a bad idea? Often yes: showing a result first, such as a partial export, gives a reason to pay.

How much data do you need to conclude? Far more than a few dozen clicks. With small numbers, talk about hypotheses, not conclusions.