How to calculate ROI using predictive or real LTV

upd 2024: The latest version of the calculator is available here. It has been updated to include revenue sharing with the publisher and changes in game growth metrics by quarter. However, the fundamental principles of calculation described below have not changed.

I offer paid consultations on mobile game launches and live operations. If you think you could use the help, just shoot me a message on LinkedIn

In the last article, we discussed how to predict LTV when your game hasn’t been released yet. Now you can estimate the approximate LTV for 7, 30, 60..180 days and use it to compare with the costs. Let’s learn how to do that.

First of all let’s decompose the theoretical ARPDAU from the last article into components which we will also could take from the GameAnalytics report. In our case ARPDAU is the same as ARPU. This will be handful for more flexible forecasting and adjustment with real data

This is genre median for 2019 year and they seem a bit low but it will go as an example for now

Next we have to decide how far we want to look. I would suggest limiting ourselves to 180 days for now, since the model is not accurate already and by increasing the period, we will reduce its accuracy even more. When we replace LTV with real values, we will make a forecast based only on the first 7–30 days and without historical data we will have a quite big chance to make an error even within 180 days.

Using cumulative method and retention values from the last article lets calculate LTV growth by month. In this step, for example, let’s imagine that we took a cohort of 10,000 people. Then we multiply LTVX by user amount and subtract the store commission (15% taken as for teams with low earnings according to new stores policies)

Please note that this is pure income without user acquisition costs.

The basic logic should be clear now. The next step good to have will be to repeat the same thing and:

1. Split the initial metrics (arpu/ret) into tiers of countries
2. Split by paid/organic traffic.
3. Split by iOS and Android

Ratio is depending on the genre of your game and UA plans but the basic idea is that Tier 1 countries usually have better paying behavior than Tier 2 countries and iOS making more money than Android. Organic/Paid traffic behavior also depends on the game and its lifetime. For some games organic shows better metrics for some paid.

For now, since we’re estimating a new project, let’s take only Android as a platform and paid traffic performs a bit better. In the first months, it may happen that your paid traffic has not yet been optimized and the results will be worse than they could be

If your game has ads you should include it to your estimations as well. To calculate ads arpdau lets define our eCPM (estimate of the revenue you receive for every thousand ad impressions) and how much ad user will see.

To find out eCPM we could take a look on appodeal article and take 10$ as an average for Tier 1 countries (let’s assume our previous organic estimates were for Tier1 country)

Amount of ads per day is up to you, lets take 5 ads as a safe option.

So ads ARPDAU will be eCPM/ads = 10$/1000* 5= 0,05$

divided by 1000 since eCPM is cost per thousand impressions

Now let’s proceed to paid traffic and put a slightly better values for paid Tier1 countries, 5000 number of users and 6$ as an average CPI for Tier1.

Joining all previous steps our 1st month cohort should look like that

Now, all that’s left is to repeat the steps for the rest of the months and add Tier2, Tier3 but by now you should be able to do it yourself.

Your total profit = sum of the all earnings - burnrate + CPI

I would put ARPU for Tier2 countries around 60% of ARPU Tier1 and regarding the Tier3 countries, it is worth putting more ads per day and leave only 10–15% of the Tier1 ARPU.

As you will get data from the soft launch you should replace draft values with the real ones. Your ltv values could be shifted depending on the game genre and user’s lifetime.

As a final step let’s take a look at how to use the first LTV results. Let’s assume we had our first cohort of 2000 people where each player already had played for one week. Get total revenue and divide it by amount of users to calculate ARPU of each day and cumulative LTV

It is important that all your players have lived the amount of days you defined as a base interval for estimations, in other case your calculations will not be accurate enough

Now we could forecast LTV using this data and approximation method from the previous article similar to retention forecasting.

Spreadsheet containing all calculations from this and previous article could be downloaded here

WHAT’S NEXT

In the next article, we will analyze:

• how to use check data for anomalies
• what are the outliers and and how to detect them

Published:

1. How to make retention model and calculate LTV for mobile game
2. How to calculate ROI and predictive LTV with the first real data
3. How to check for data anomalies and outliers
4. Classification of users. KNN method
5. How to check the representativeness of the data sample

Upcoming:

1. How to identify the correlation between events. An example based on user’s actions on the first day and their impact on retention
2. What types of data do we usually work with in mobile games
3. Practical examples of the choice of statistical criteria
4. Bootstrap method. How to identify statistical significance on a limited date. What are its advantages
5. How to reduce the date accumulation time in a/b tests