How do you deal with whales in mobile game LTV calculations and predictions? From a statistical point of view, they are extreme cases or outliers. Do you recommend including them in the calculations even though they significantly distort the result? What do you suggest when 50% of revenue comes from whales?
Answer from our Expert:
This is a very relevant topic in free-to-play (F2P) games. Of course, it depends on your definition of a “whale” but it is often the case that most of the revenue comes from these big spenders. Let’s consider for example that 2% of players decide to make a purchase. That means 100% of revenue comes from just 2% of the playerbase. But from this 2%, most of the payers buy only one offer for a few dollars (usually a starter offer) and only a fraction of these payers spend significantly more. This leads to an extremely skewed distribution of revenue per player.
So, simply put, we cannot work with it as we do in basic statistics and disregard whales as outliers. That would considerably decrease our LTV (and ROAS) estimates.
Every time you calculate LTV for mobile games, you need to be very careful of the variance. When you predict the LTV of cohorts, make sure that each cohort has a reasonable sample size of spending players. You can have 10 000 players in a cohort but if there are only several payers (players who contribute to the LTV), you can expect a huge variance of error. You can also identify this with big irregular “jumps” in cohort LTV curves (as shown in the images below). We are looking for cohort sizes with a relatively stable LTV curve.
This cohort is definitely too small to have a reliable prediction.
This is much better and LTV predictions can be made with higher certainty.
It is important to note that this is mainly an issue for games that monetize heavily by in-app purchases. If you monetize more by ads, your LTV curves should be smooth because the sample of players that contribute to your revenue is much higher, making calculating your mobile games LTV much more accurate.
The only situation where it makes sense to exclude whales as outliers from the measurements is a monetization A/B test. If you compare ARPU between groups, the absolute value of the metrics is not as important as their difference. In this case, few whales can distort the true difference between the groups.
Viktor Gregor, Senior Data Scientist