NewThe guide: 11 paywall examples

7 minutes

Mobile app ARPU: definition, calculation, and denominator

A mobile app's ARPU is a period's revenue divided by its users. Install, active, or paying: the same money does not produce the same number.

Clément Parot

Mobile app ARPU: definition, calculation, and denominator

Article summary

  • ARPU divides a period's revenue by the users of the same period.
  • Three denominators coexist: the install, the active user, and the paying user. The third is ARPPU.
  • In the article's example, €20,000 yields €2 per install, €0.50 per active user, and €25 per paying user.
  • At Mukiz, a client of ours, an iOS subscriber brings in three times more after the trial, the daily pass, and the monthly plan are removed. Revenue is up 98% over the same three months, with the team's ads.
  • To keep a test, firstapp reads revenue at thirty days per install, with the plan mix alongside.

Summarize this article with

A mobile app's ARPU is the revenue of a period divided by the users of that period. The acronym comes from English: average revenue per user. What changes the figure is who you put underneath: an install, an active person, or someone who paid.

What is a mobile app's ARPU?

ARPU measures what a user brings in, on average, over a window you have chosen. That window can be a calendar month, or the thirty days that follow the install. It is part of the figure: without it, you do not know which period you are talking about.

firstapp reads it on subscription apps that are already live, alongside conversion, trial-to-paid, and realized LTV. How to use it is detailed on Growth.

Two apps can announce the same ARPU and not be talking about the same money. One divides by installs, the other by paying users. Both calculations are correct. They do not describe the same thing.

Newsletter

Once a week.

Last issue

We opened 10 apps to see how they sell a subscription.

+534 people read us each week

How do you calculate a mobile app's ARPU?

The formula fits on one line.

ARPU = revenue for the period / users in the same period

You add up the subscriptions and purchases for the period, then you remove refunds. The users are those in the same window. Change the revenue, or the base, and the result moves.

💡 An example, not a measurement. An app takes in €20,000 in the month. This is not a firstapp figure. It is the same revenue, split three ways.

DenominatorBaseARPU
Installs in the month10,000€2
Active people40,000€0.50
Paying users800€25

The bank account has not moved: it is still €20,000. Yet ARPU per paying user, €25, is fifty times the ARPU per active user, €0.50. All three rows are correct, and they describe three different populations.

ARPU and ARPPU: what is the difference?

ARPPU, average revenue per paying user, is the same calculation once free users are removed. You only divide by the people who have paid at least once.

ARPU looks at the whole base. ARPPU looks only at those who took out a card. It is therefore always higher, and the gap between the two says what share of the base actually pays.

In the example above, €0.50 per active user and €25 per paying user: the money is concentrated on 800 people, it is not spread across the 40,000 active users.

Why ARPU can rise without revenue rising

Two movements make ARPU rise while the money itself has stayed put.

The first is a base that empties out. The following month, still €20,000, but only 20,000 active users remain. ARPU per active user goes from €0.50 to €1. It has doubled, and revenue has not moved: the denominator has shrunk.

Active usersRevenueARPU per active user
Month 140,000€20,000€0.50
Month 220,000€20,000€1

The second is the mix of plans. An annual payment pays several months at once, in the month of the sale. That month's ARPU rises. In the following months, that person does not pay again, and ARPU comes back down. The figure followed the cash collected. It did not say what the cohort will be worth. Weekly and annual are detailed in weekly or annual.

A price test produces the same kind of illusion. Conversion falls, revenue per paying user rises, and the month's ARPU looks better. Until the plan mix is set alongside it, you do not know whether the app made money or collected the year in one go. The pricing frame is in the pricing strategy.

Mukiz: the denominator on a real app

On Mukiz, a client of ours, five prices sat on the same screen: the trial, the daily pass, the monthly, the weekly, and the annual. We remove the trial, the daily pass, and the monthly. The annual and the weekly remain. As many people subscribe as before.

After three months, an iOS subscriber brings in three times more. Over the same three months, revenue is up 98%. The case separates the two causes: the ×3 comes from the offer, and it is the team's ads that brought the people in.

If you only read the +98%, you attribute the gain to the ads. If you only read the ×3, you forget that the base, alongside, has changed. And what a subscriber brings in is not an ARPU per install: it is revenue per paying user, over time. Both numbers are true. They do not answer the same question.

Which ARPU should you track on a subscription app?

You can produce all three. You keep only one for decisions.

ARPU per install, at thirty days, is the one you set next to a decision. ARPU per active user is for seeing whether the base is emptying out. ARPU per paying user is for seeing the basket. Calendar-month ARPU moves with the season and with the mix: it is for monitoring, not for declaring a winner.

The reading order is the same from one test to the next. First the trial rate, which is easy to inflate. Then trial-to-paid, which says whether the offer promised too much. Then revenue retained at thirty days, per install. Finally realized LTV, for apps that can wait. That order is on the Method page.

ARPU remains a period. LTV follows the same user for longer. The projection formula, and the market median, are in LTV and CAC of a mobile app. The list of surrounding indicators is in mobile app metrics.

The mistakes that distort ARPU

Forgetting the denominator. "€5 of ARPU" compares to nothing. You have to say at the same time whether it is per install, per active user, or per paying user, and over which window.

Taking a rise for a gain. ARPU rises when revenue rises and the base stays the same. It also rises when free users disappear, or when an annual plan pays several months at once.

Calling a test on the month's ARPU. You iterate on the trial rate. You keep the variant only after revenue retained at thirty days. If the trial winner loses on that revenue, the offer promised too much.

Comparing a subscription app's ARPU with that of an ad-funded game. One divides subscriptions by installs. The other divides advertising by daily active users. The name is the same. The base is not.

FAQ

What is a mobile app's ARPU?

ARPU is the revenue of a period divided by the users of that period. The denominator is an install, an active person, or a paying user. The window is a month, or the thirty days that follow the install.

How do you calculate ARPU?

You add up the revenue for the period, once refunds are removed, and you divide by the users of the same period. ARPU = revenue for the period / users in the same period. Changing the denominator changes the result. In the article's example, not a firstapp measurement, €20,000 makes €2 per install, €0.50 per active user, and €25 per paying user.

What is the difference between ARPU and ARPPU?

ARPU divides by the whole base. ARPPU divides only by those who have paid at least once. ARPPU is higher, and the gap between the two says what share of the base pays.

Does a rise in ARPU mean the app is making more?

Yes, when revenue rises and the denominator stays the same. No, when free users leave the count, or when an annual plan collects several months at once. You set revenue alongside it, and the plan mix.

Which ARPU should you track to keep a test?

Revenue retained at thirty days, per install. Calendar-month ARPU is for monitoring: it moves with the mix and with the season. You reread it by channel and by country before keeping a variant.

Newsletter

Once a week.

Last issue

We opened 10 apps to see how they sell a subscription.

+534 people read us each week