In Data analysis, Games Analytics

The games industry is awash with buzz around the term gaming analytics.  All publishers and developers recognise that analytics has to be a key part of their strategy and as games distribution moves more and more online, to truly maximize profit and positive feedback from players,  revenues need to be secured by first understanding player behaviors.

But there is general confusion on what is meant by ‘analytics’ and what sort of analytics to invest in to maximise returns.

The main types are:

Metrics and Dashboards

The industry has devoted considerable thinking and energy into defining key sets of metrics.  Stickiness, K-Factor, ARPDU and other more exotic terms are often used to gain insight into game performance.  But rarely do they tell you much about the different types of players in the game and how they are behaving.  And although metrics will tell you what your retention rate is; they will not tell you how to improve it.  Therefore dashboards are an important monitoring tool but other solutions are required to provide the insight to change the game and move the metrics.

Funnel Analysis

Funnel analysis starts to take a player view by defining milestone events in the game and measuring how many players move through each milestone.  This is a simplified version of a statistical methodology called survival analysis which is a technique to understand and interpret complex defection behaviours.

Funnel analysis will potentially uncover bottle necks in the game but there are two issues.

One – it’s trial and error; the user has to guess which event milestones might be important; in survival analysis each event is analysed and the highest impact events isolated.

Two – most software implementations of funnel analysis do not work on player cohorts so the results have to be used with caution. However funnel analysis can be a useful tool if properly implemented.

Multivariate Analysis and Predictive Modelling

This is the most sophisticated way of analyzing event data at player level.

Using statistical techniques it is possible to understand the patterns of game play and event combinations that players undertake.  Multivariate means assessing many events at the same time rather than looking at events individually.  Furthermore these behaviours can be assessed to determine the likelihood of players leaving the game; or making a purchase.

The power of this approach is that it is possible to focus on the behaviours that are predictive of defection or predictive of purchase so that game design and in-game messaging strategies can be devised that are relevant for individual users.  This is a direct way of increasing retention or monetization of games by being relevant to the individual user experience.

So there are different forms of analytics available but the ultimate way of increasing the impact of analytics on revenues is to use all three in combination to provide true insight on how the players are experiencing the game.

Recent Posts

Start typing and press Enter to search