Gamasutra: The Art & Business of Making Gamesspacer
View All     RSS
September 18, 2014
arrowPress Releases
September 18, 2014
PR Newswire
View All





If you enjoy reading this site, you might also want to check out these UBM Tech sites:


 
The Importance of Tracking by Country
by Aaron Isaksen on 11/28/12 10:38:00 am   Expert Blogs   Featured Blogs

The following blog post, unless otherwise noted, was written by a member of Gamasutra’s community.
The thoughts and opinions expressed are those of the writer and not Gamasutra or its parent company.

 

One of the first things we noticed the day Chip Chain launched on iOS was that each country acts very differently, so it's essential that you track which country a user comes from, or else your data can easily be corrupted.  And it's equally important to look at these differences when deciding to make a change in your app. Luckily, I had put this information in the tracker data for purposes of figuring out revenue from each country, since this is reported in the local currency, so all our tracking events come with a country identifier.  I expect that almost all trackers already do this (for example Flurry does), and you should not use a tracker that doesn't allow easy separation by country and other factors.

When analyzing your data, this is called a "segment" which is a subset of the complete set of users.  You can do segments based on all sorts of things like age, gender, spending habits, country, language, etc.  But for this discussion, I'll focus on two segments, one for all users from the US and one for all users from China, since those are the top two markets for Chip Chain.

The US is home to 26.57% of our users, and China is home to 31.08% of our users.  When we look at how much revenue I get from IAP from each country, since Apple reports this to us on a daily basis, we get significantly more revenue from the US…in fact it's currently 7.16x more total revenue from US, even though we have more users in China, and on a per user average, it's 8.17x times more revenue from an "average" US user.  This is pretty critical if you are looking at stats like ARPU (average revenue per user) since the numbers are so different depending if you include China or US.  Attitudes to paid content and piracy are certainly different in these two regions, but we can look at more data and see there is still more going on than just differences in attitudes to paying for IAP.

For example, we can look at the number of sessions a user has played the game.  A single session is basically if someone opens the game, plays a couple of levels, then closes the game and doesn't come back for a while.  If the user returns within a short period of time, lets say 10 minutes, it still counts as the same session.  I see that 40.6% of US players have come back for 10+ sessions, while only 31.7% of Chinese players have come back for 10+ sessions.   I don't know what it is, but there is some reason the game isn't yet as compelling for Chinese users as US users.

Perhaps the difference is just due to local tastes: gamers in different parts of the world prefer different types of games, so it's possible that a game is niche in one country while it's mainstream in another. Local tastes can also explain why there are differences in the amount of time played or when the games are played.  Or it could be localization issues - Chip Chain is translated into 9 languages, including Simplified Chinese, but perhaps it has mistranslations or confusing translations on it's first release.  Over time, we should build up enough fans that would tell us when a translation is incorrect so that we can tune this up.  Furthermore, In our first 2 builds we had some bugs that only showed up in certain languages, so this could affect retention for those specific languages, although this didn't occur for Chinese or English.

The important lesson is that you need to pay attention to each region differently, and only lump countries together if you notice similar patterns of behavior.


Related Jobs

Vicarious Visions / Activision
Vicarious Visions / Activision — Albany, New York, United States
[09.18.14]

Character Artist-Vicarious Visions
Cloud Imperium Games
Cloud Imperium Games — Austin, Texas, United States
[09.17.14]

Lead Network Engineer
Cloud Imperium Games
Cloud Imperium Games — Santa Monica, California, United States
[09.17.14]

Animation Programmer
Cloud Imperium Games
Cloud Imperium Games — SANTA MONICA, California, United States
[09.17.14]

Art Director






Comments


Benjamin Delacour
profile image
Is the US consistent across states? Have you tried a grouping algorithm across all user data?

Aaron Isaksen
profile image
I don't have access to which state the user came from, just the country. We don't use GPS location in our game. Can you elaborate by what you mean by "grouping algorithm"?

Benjamin Delacour
profile image
I mean "data mining" or "cluster analysis" software for automatically determining patterns in large amounts of data. Some patterns are difficult to detect by guessing traits or looking at statistics.

Aaron Isaksen
profile image
Haven't used any sophisticated data mining software. Would you recommend something?

Yannick Bourquin
profile image
(Please excuse my English, I'm not a native speaker)

This is a very interesting point of view that highly constrats with my personal experience. I'm usually not a big fan of analysis on a per country basis (except maybe for key figures: DAU, revenue...).

The last time we attempted such a study on customers habits, we ended up with differences that we could not interpret. We noticed for example that German players tend to buy a particular set of items more than other users. But we don't have a clue why it is so and we can't recommand any line of action based on this observation. So, the study was a failure.

The only two decisions that we take on a per country basis is: Should we keep translating into this particular language? Should we keep on acquiring users in this particular country?

I agree that it can also raise alarm on translation issues or language specific bugs, although it will seldom appear as such in the data.


none
 
Comment: