What's lurking behind LTV?
In the highly competitive world of mobile gaming, understanding the lifetime value (LTV) of your users can be a challenging task, but once you have it down, you will be one step ahead of the game in knowing which marketing channels best deserve your attention and marketing budget.
Simply put, the LTV of a user is the revenue you can expect to gain from that user over the course of the users’ engagement in your app. This is not a static number. Because the LTV changes and increases over the lifetime of a user, analytics are essential for knowing what is really going on.
There are two main methods for projecting LTV, either based on modeling after certain events (suitable for when your data set is limited), or predictively, which is based on preliminary data captured at certain time points. Whichever way you go, you can leverage your LTV by identifying which marketing channels are giving you the best results, and invest your time and money in these channels.
Let’s imagine that you have a free to play gaming app. Unlike freemium or premium apps, your LTV varies rather than being fixed to certain points, and is therefore probably one of your core metrics. You’re looking partly to upsell, so to boost the LTV of existing users, and partly to acquire the users that are most likely to engage with your content.
The easiest and most intuitive way of working this out is by calculating the average LTV from your users over a longer period of time and watching that metric as a single, straightforward KPI. If it moves upwards, good, and if downwards, bad. Easy.
We did this together with one of our gaming clients on their data for a couple of months. Then we ran another analysis, based on cohort analysis, to get a firmer grasp of the big picture.
Looking at the first metric, everything looks good - the graph looks pretty much like you expect it would if the metric had been effectively and consistently optimized. When you mix in the second metric, though, the results are startling:
This graph shows the above LTV versus the cohorted, 14-day average revenue per user (ARPU). We see that the individual, early LTV is up to 2.5 times the extent of the maximum average LTV – and that it goes pretty far down, too.
The plateau wasn’t LTV being optimized! The revenue of incoming users previously collapsed, meaning that new cohorts had a dramatically lower LTVs than users who installed a previous iteration of the app. This crucial piece of information was completely fogged up by a flawed KPI in the first analysis.
This is how an effectively implemented cohort analysis allows you to calculate LTV at a much earlier stage and receive quicker, crucial feedback on your optimization, where a standard, retrospective average LTV may even hurt your analysis.