Posts Tagged ‘games’

Sony Online Entertainment leverages Sonamine Predictive Player Scores

Wednesday, January 2nd, 2013 by Nick

Happy new year everyone!

There is a lot of industry interest in using data analytics in games.  Looking at the gamasutra or linkedin job board today, I see that Ubisoft, 2K games, MachineZone, Ngmoco, Microsoft, Z2live, Activision etc are all looking for data analysts.  Especially after it was revealed that the Democrats used a predictive scoring algorithm to allocate scare volunteer resources to get-out-the-vote, interest in using predictive analytics has never been higher, or more hyped.

But really, how do you actually use predictive scores in a day-to-day productionized manner to increase revenues and retention?  Well, one of Sonamine’s customer is Sony Online Entertainment, an industry leading MMO developer.  They have successfully integrated Sonamine predictive player scores into a full fledged player relationship management program that encompasses different player touch points, a communications calendar and different offers.  SOE presented their story at GDC Online in Austin.  (skip to link at bottom to get presentation).  Here are some of their lessons:

  • Develop a player relationship management program, with calendar of communications and offers.
  • Have a system that allows targeting to different sub-segments of players with appropriate messages.
  • Use Sonamine predictive scores to create different player segments.
  • Take the long view, it is a journey.
  • Do not underestimate the resources needed to pull this off.

Fill in a short form and download the Sony GDC presentation here.

How will facebook feed changes affect social games?

Friday, September 23rd, 2011 by Nick

Facebook made some changes recently that will have big ramifications for social game developers.

For starters, friends can now be grouped into different lists, and posts from close friends will figure more prominently on the feed.   In the past, when a player, say Alice, of a facebook game published an announcement like “achieved level 10″, this post would be seen by all her friends.  Now given, this new listing capability, friends who have listed Alice as close friends are much more likely to see this post.   Some “acquaintances” of Alice may never see this game post at all.  While this list feature will help users reduce the clutter in their news feed, it also means that game developers have fewer opportunities to reach out to friends of their players.

Secondly, Facebook will reorganize the news feeds based on whether the user has been back frequently.  If a user has not been back for a while, Facebook will try to show the most important feeds.  Some type of relevance calculation will be taking place to determine whether a game published post is appropriate as  “important” news.  This feature will, in theory, reduce clutter and highlight the most useful feeds.  Alas, for game developers, it again means that fewer users will be seeing the game published feeds of the players.

The net result is that game published feeds will be viewed by fewer users than before.  This organic channel of player acquisition will be less efficient and thus the overall cost-of-customer acquisition is going to rise.  In a sense the changes are part of a larger trend that had already started : the loss of the virality channel on Facebook.  Game developers have been adapting to it, and now they will have to be further vigilant to make sure that their game published feeds are more relevant and appropriate.

What to do about it?

How Facebook will determine the relevance of various posts is not clear, although it is intuitive to think that a game published feeds that has many click-through will be “more” relevant.  Game developers will have to play a balancing act between allowing too many posts and getting more click throughs.  The first step is to understand what your click-through rates are from your game published posts.   Some games are already tracking these events, but if you are not, now is a good time to start.

When your game publishes a feed, make sure you include query parameters in the link to identify the referrer and other information about when the post was made.   Additionally,  count the number of posts by each user.  This will allow you to identify who posted what and when.   A click through will pass the query parameters into your game.  Make sure to grab these query parameters and store it as a referrer for each click through.  Now you can identify how many  clicks resulted from individual posts.

Secondly, it would be now useful to track the social graph of your players.  The goal is to see players with specific types of social graphs are better at getting click throughs.  Some aspects to understand include number of friends, number of friends in various lists if that is available, the amount of messages and invitations sent by this user within the game.

The next step is to predict which players are most successful at getting click throughs.  These are influential players that can get friends on-board.  Shameless plug: Sonamine InfluenceSoon helps you do just that!

Once you have a handle of which players are good at getting click throughs, you must make a game design decision on what you will with this information.  Some options include making it easier to post for these influencers, or vice versa, making it harder for non-influencers to post.  That is a game design question that will comprise the balancing act.

Good luck!