Game Balancing: Data-driven Approach
Game success assessment
Igor Klyukin, COO Pixonic
As an analyst, I often face the task of comparing different projects or one and the same project at different moments in time. I’d like to tell you how it is done at Pixonic. Of course, some would say: “Dough is the most objective parameter of all. Income is the best basis for comparison.” However, income is a result of work not only associated with creation of a project, but its promotion as well. And the assessment is mostly required for the projects launched recently, far from completion.
First, I’d like to explain why game comparison using ARPDAU is wrong. Many analysts use this parameter to assess the general picture, since it reflects income per active user without consideration of the amount of active users, which could be few, many or even a lot. Besides, the said parameter does consider the average check (ARPPU ), and the percentage of paying users, ignoring, however, the amount of installations turned to loyal users. Therefore a high ARPDAU project may actually turn out to be worse than one with low ARPDAU. That’s why we first review the income per installation (ARPI).
The figure visually shows how among two projects with the same amount of installations, the one with worse income and ARPI results turns out to have a higher ARPDAU value.
Second, I’d like to mention that during assessment, it is important to view parameters by cohorts, including the ARPI. Cohorts are groups of users broken down by registration dates and parameters for such groups are to be viewed with reference to the installation date. For instance, 7 days after the first launch. For user cohorts who installed the game on June 1st and 10th, parameters on the 7th day of the game should be measured on June 8th and 18th, accordingly.
Thus we can be sure that advancement of all players in the game is approximately the same by the time parameter values are taken.
Third, it is important to mention that parting from the parameter values we provide, the developer has specific decisions to make. Therefore we measure not only ARPI by cohorts, but four other important parameters as well: player Retention on the 1st/7th day, amount of installations turned into pay users and average revenue per pay user.
Basing on the values of those parameters we make decisions regarding the fate of the project and specify which aspects of the game require a focused aspect of the developers. Here is our algorithm of processing these parameters.
Sign in to comment