Mobile/Free-to-Play Forecasting...

How to approach it when you have no data to go on and so much for Sunday-morning newsletters!

Jan 16, 2024

Greetings and happy 2024 (we can still do that right?!).

This year got off to a good start for me. I finally managed to get the free-to-play/mobile game forecasting tool I’ve been working on for months finished and published. You can check it out and request access here.

Now, there are lots of tools out there which seek to do this kind of modelling already, so you might be wondering why I built this...

How a lot of existing models work:

Most of the ones I’ve seen are based on metrics like:

D1, D7, D14, D30, D90, D180 retention - i.e. the % of people who come back to the game/app those number days after install.
Daily Active Users (DAU) which you can model from the above by factoring in User Acquisition (i.e. how many users you expect to bring into the game from advertising) and other marketing efforts.
Average Revenue Per Daily Active User (ARPDAU) which you can then multiply by DAU to calculate daily revenue.

The limitations of that approach:

These kinds of models are hugely useful during soft launch and live operations when you are able to look at, for example, D30 retention and estimate what the likely effect of increasing it by 20% might be on revenue. However, there are a number of problems with using these kinds of models in the early stages of a game development process:

These metrics can vary hugely from title to title, and until you hit soft launch/live operations with yours, you can only really guess at what they might be.
Therefore, the variables you’re basing your model on are going to be prone to uncertainty.
The greater the number of uncertain variables you base a model on, the more scope for error there is in it, as each level of uncertainty magnifies the last.
Metrics like ARPDAU and Retention aren’t hugely helpful for focusing the mind of a game designer on what they need to achieve. Metrics like Average Revenue per User (ARPU) or even Per Paying User (ARPPU) are more useful - particularly if you can split them into cohorts of low, medium, and high-spending users.

So, my model is set to address that! What I’ve done is taken Revenue Per Download data for live titles (which is a reasonable proxy for ARPU/LTV) from Appmagic and used that, combined with Cost Per Install benchmarks/estimates for different types of titles, to create a simpler approach which - hopefully - creates a clearer estimation and set of targets for the team.

More information on the approach can be found on the introductory page & accompanying notes. It’s important to note that this is only designed as a starting point, but hopefully, it’s a good one!

And… no more Sunday newsletters!

I’ve decided to abandon the Sunday morning release plan for a couple of reasons:

You don’t want to read stuff like this then, do you?
I’ve realised I only want to send out newsletters when I have something to say!

There are much better newsletters out there if you want regular updates - I recommend checking out Mobilegamer.biz in particular - Neil provides a great round-up every week.