So, if you read any of my Game Room posts you might wonder why i do so many posts on sales estimates and the like. The simple answer is that for some reason Game Room brings out that side in me. Maybe it is that it is a reminder of a simpler time where possibility still existed.
The long answer goes back to my days in college. i have a degree in Business Administration (Finance) and Economics (Managerial Economics). What i really would have liked to do after college to be a corporate financial analyst (working mainly with cost of capital type stuff). For some reason, things like NPV, WACC, IRR, time value of money, etc. seemed to click for me (seemed easy to understand), however i was working with simplified scenarios and theory mostly - because that is mostly what college is.
When Game Room added leaderboards i took more of an interest in the results and behind the scenes type stuff. This started with my posts on the official Game Room forum with my post series The Business of Game Room (which i re-posted here). In that series i was trying to look at Game Room like a business analyst (arguably not my best class in college). i was trying to figure out how it might operate (using different aspects). Eventually, that evolved into stat tracking/recording. i have always been interested in playing around and interpreting stats (though i can't stand hypothesis testing), however, i am more than a little out of practice. Sometimes partial concepts pop into my head though.
Then i turned it into a sales estimate quest. What i decided to use were simplistic methods (at least in my mind). i started by estimating that real sales were 75% of leaderboard data - mostly because i was thinking in terms of multiple accounts on one console and the pay for one play option. After looking on a sight that estimated XBLA games and seeing that they estimated that something like 90% of leaderboards were sales to individual accounts i changed my assumptions. i now use a conservative point estimate of 87.5%. That breaks down to an estimate that 85% of leaderboard entries are unique purchasers and 2.5% added because of the way the leaderboards are set up (essentially that real leaderboard figures are 2.5% higher than the plateau level shown). i am seriously considering increasing my point estimate to 90%.
Of course i wasn't content to let thing stand at those conclusions, so i added another wrinkle. That wrinkle was irr. irr is actually my attempt to estimate what the leaderboard numbers are (putting aside the 2.5% estimate) based off of the average leaderboard position divided by the number of weeks a game has been on the service. After that i kind of messed up offshoot of NPV and use a multiplier to create weekly estimates, which i then sum. The first numerical entry in my list of weeks will be the first week's observed leaderboard information. i messed up the timing on data collection (should have been Tuesdays instead of Thursdays/Fridays), so i can't get a handle off of second week declines and things like that.
Maybe i will come up with a totally convoluted metric in the future. i am actually interested in how much i should discount sales in subsequent weeks - but i'm not sure where to start with that (as i don't have the data points i would like).
No comments:
Post a Comment