In Data analysis, Games Analytics

The games industry is changing.  The retail model is fading and being replaced by games that are consumed online.  Instead of paying $75 upfront for the latest console release, games are now available across a wide range of connected devices and often they are free to play.  Even the next-gen consoles from Sony & Microsoft reinforce this change with an emphasis on online games not boxed product.

This has brought about some fundamental changes in how games are designed and developed.  Firstly the development process is no longer focused on a waterfall approach that delivers the games into stores to a specific deadline (generally in time for the Holidays).  Rather games are released and developed on an ongoing basis with new features and content being constantly refreshed to retain and engage players.

The free-to-play model was devised to maximise the number of players that accessed online games.  But this has meant that it is vital that games engage players and create an environment where a proportion of players will spend real money in the game for a faster or broader experience.

Therefore the games industry, in less than five years, has gone from knowing very little about its customers to needing understand players’ experiences and build responsive environments.  Games should be fun; first and foremost.  The creative process is fundamental.  But it is now possible and vital to supplement this expertise by using analytics to understand player behaviours and put this insight at the heart of how games are produced.  Customer Relationship Management, often the focus in traditional sectors such as finance and retail, becomes Player Relationship  Management in the virtual world of the game.

The prize for doing this well is huge.  Companies like Supercell  and  generate $millions per year in revenues and have player bases well into 10s of millions.

deltaDNA is leading the charge in the games industry to get more intimate with its players.  But there are some interesting challenges to overcome for analysts.

The data volumes are vast. It is possible to collect every action by every player.  Each time a player starts a mission, buys an item, invites a friend can be collected within the anonymous environment of the game.  Also the playing behaviours are complex and there is a real data reduction problem to understanding what is going in with different segments of players.

Also time is a critical factor; the game is being constantly developed so careful cohorting of playing groups is essential for analytics to deliver robust outputs.

But aside from these challenges, by using varieties of existing statistical techniques combined with the development of new ones, it is possible to characterise different playing styles and start to put these insights to good use in how the game is developed.

For example, we recently worked on a Deathmatch tournament style multiplayer game available on iOS and Android. It was clear from the game performance metrics that retention rates were low and impacting  the potential revenues.

The more traditional route would be for the designers to get together and brainstorm possible solution before committing one to the development team.  Intuition without insight.

By looking at various factors such as competency, momentum, rewards and intensity of game play, we were able to create strong behavioural segments that showed fundamental differences in how contrasting playing styles were experiencing the game.

For example, one segment which we called Strategic Elite demonstrated very high skill levels, high win rates and they conserved their resources.  However this ‘Snipper’ playing style was being under-rewarded by the game, these players were getting frustrated and leaving.  In contrast our ‘Kamikaze Killers’ segment wasted their resources, spent much of their time out of the tournaments whilst healing but were being over-rewarded by the game.  This meant that they didn’t monetize as they had plenty of in-game currency to replace weapons and ammo as often as they liked.

By taking a player centric view of the issue, suddenly it showed that the in-game economy was badly balanced, and that by restructuring the rewards it would be possible to 1) convince the ‘Snippers’ to engage with the game, and 2) encourage the ‘Kamikaze Killers’ to spend some money to enhance their experience.

This is the power that analytics can now deliver into game development – taking out some of the guess work so that important decisions are based on actual player behaviours.

But more than that it is also possible to take this approach to its natural conclusion and target individual players with real time in-game messages that encourage engagement and retention – making games more successful and sustaining the industry in the new and challenging world of free-to-play.

And so, as we know, it’s not Intuition vs Analytics; it’s Intuition and Analytics with analytics taking its rightful place alongside outstanding creative ideas, great game design and expert development to ensure that the guesswork is taken out of the development process and games become genuinely responsive to players.

Recent Posts

Start typing and press Enter to search