With the kind permission of Alexey Pontyakov, mobile Games analyst Mail.Ru Group, we will repost with habrahabr.ru his material is about how to correctly calculate the average revenue from the user.
ARPU – average revenue per user, or “average revenue per user”. If you think that this indicator is calculated by dividing all revenue by all installations in time from the moment of release and to date, then this article is for you.
Many, due to lack of knowledge, and maybe because of the desire to simplify their work, are trying to calculate one of the most important parameters of the game according to the above scheme. And this is wrong. Especially when the LifeTime (the lifetime of the player in the project) of the user is less than the period in question by 2 times or less.
For clarity, I suggest you familiarize yourself with the table:
The first column is the cohorts of users, united by a common feature “registration date” (downloading the application and launching it by the user). The registration dates were broken down by weeks.
The second column is the number of registrations for this period. It is very important: take registrations from your database or from the statsystem, which considers the user’s entry into the game for registration, and not just downloading an application from the market (in the article I mean only Google Play or Apple AppStore by market). Keep in mind that in the market, “downloading” is considered installing an application on a device, not launching it.
And also, with a certain error, we are forced to limit the user to one device (1 device is 1 user); thus, the user can install and launch the offer from two devices, and we will assume that these are two users. Unless, of course, we use the services of our own database and force the user to somehow create an account (or log in to social networks) in the game and link the device to it.
1 – 10 weeks – weekly earnings.
On the first day of the ninth week, we were given a task – to calculate how much the user brought us in the first eight weeks.
Let’s imagine that the average user lifetime in our application is exactly three weeks. Well, let’s go down a simple path, solve the problem “head-on”.
In the weeks from the first to the eighth inclusive, 68 thousand users have registered in our application. We earned $51,680 in eight weeks. Divide the earnings by the number of registered users and get ARPU = $0.76.
BUT! Our main mistake is that, knowing the average user lifetime of three weeks, we did not give the opportunity to pay users who registered from the second day of the sixth week to the last day of the eighth. Users from the first to the fifth week had more than 21 days to make a payment, whereas these had 20 or less.
Therefore, we can get the correct data for calculation from registrations in 1-5 weeks.
So, let’s try to calculate ARPU more accurately. We take earnings from installations that came in the first five weeks to date, and the number of installations that came in these five weeks. Divide one by the other ($32 150 / 35 000 ), we get ARPU ~ $0.92. This is $0.16 or 21% more ARPU from the first option! Consequently, with the same volumes of installations, our revenue will be 21% more. With large volumes of advertising campaigns, this 21% can seriously affect the marketing strategy.
Let’s imagine that the project was available for download only for 10 weeks. Taking into account the lifetime of users and the fact that each subsequent week each group of users paid the same part from the first day (the coefficient decreased with weeks), the final ARPU after the project “died” will be $0.93. This ARPU is the most correct. On the first way we got an intermediate value, which is $ 0.76, on the second – $0.92.
For such parameters, an error of more than 10% is unacceptable. A living example is PopCap with their Plants vs Zombies 2 gained 25 million users in two weeks. Revenue from ARPU $0.76 = $19,000,000. Revenue from ARPU $0.92 = $23,000,000. Some lost 16 cents turned into an uncounted $4 million. Not bad, right?
In conclusion, I will add that we can consider the LT of a separately paying user (ARPPU), but for this we need to know the Paying Conversion, which strongly depends on the sources (quality) of traffic and monetization methods.
ARPPU – Average Revenue Per Paying User. Average revenue per paying user. It can be expressed in terms of ARPPU=ARPU/PC. For example, Candy Crush Saga has ARPPU=$40, Paying Conversion=8%, hence ARPU=$40*8%/100%=$3,2.
PC — Paying Conversion. Conversion to a paying user. If the game was installed by 100 users, and the conversion rate is 2%, then we have two users in the game who have paid at least once.
That’s it, thank you for your attention! I am ready to answer your questions in the comments.
PS: If you suddenly decided to publish data on average earnings from a user – forget about the commission of the markets! Competitors should be jealous!