Recently I got asked to do a UI study for one of the major mobile ad providers. I didn’t read the NDA very closely so I will remain very vague about most aspects of that conversation, but the topic of LTV did come up. I explained what I currently do and they mentioned how tedious it sounded. As I mentioned in a previous blog entry, I don’t currently use any analytics provider that automatically calculates ad revenue LTV. Mostly because I can’t afford it, don’t want to add additional bulk to my apps, or the free ones do not integrate with Appodeal (my monetization provider).
So what exactly did I do until about a month ago for LTV? I would use calculator number 2 from this 2006 blog post from people at Soomla. I had a Google Sheet with the formula and I would just plug in the retention rate. As the blog post explains this would give me average days played for the particular group (cohort) and then I would multiply that value with the average revenue per daily average user (ARPDAU). Appodeal provides this value in their reporting tool. I would group it by country because for the most part, it is the most convenient grouping for the types of advertising that I do (probably a topic for another blog post).
After a lengthy conversation with the people from the UI study group, I realized that I didn’t really need the retention formula to guestimate the number of days that an average person returns to my game. I have the data readily available in Gamesparks (what I use as the backend for most of my games).
A while back I decided to reward people that came back to my game on certain days. This was based on the fact that I lose most of my players on the 2nd and the 5th day. To that end, I record on Gamesparks the number of days you have played and then on the 1st and 4th day, I let you know that if you show up tomorrow you’ll get a prize. It has not been very effective, but it was an idea.
In any case, I have 6 months of active user data to run my LTV calculations. The first thing that struck me when comparing my old calculations with my new ones is how off the estimated days played were for some countries. Since I have not been inspired to write blog post lately, I have decided to show anybody who reads my blog how I am currently calculating LTV with data from Appodeal and Gamespark.
So consider this one the introduction. The next couple or few blogs post will be all about LTV (how exciting). And as a teaser here is what the graph of my current spreadsheet looks like: