Page 514 -

Chapter 12 Enhancing Decision Making 513

A small start-up company can deploy Vertica on a audience, and to generate revenue by selling virtual
single node, adding new nodes as needed. goods in game. The idea of virtual goods has been
Vertica’s data warehouse is columnar, which around for years, most notably in Second Life and
means that data are stored in columns instead of other virtual worlds, where users can buy apparel
rows. This allows Zynga’s data to be more tightly and accessories for their avatars. But Zynga’s atten-
compressed, at a rate of 10 to 1 (10 terabytes of data tion to detail and ability to glean important informa-
become 1 terabyte of compressed data). Vertica’s data tion from countless terabytes of data generated by its
warehouse is able to work with this compressed data, users on a daily basis has set it apart.
which improves performance by reducing processor For example, product managers in Zynga’s
demands, memory, and disk input/output at process- FishVille Facebook game discovered that players
ing time. Traditional database management systems bought a certain type of fish in game, the translucent
can’t work with compressed data. As a result, Zynga anglerfish, more frequently than the rest. Zynga
achieves rates of performance that are 50 to 100 began offering fish similar to the anglerfish for about
times faster than the data warehouses used by other $3 apiece, and FishVille players responded by buy-
companies. ing many more fish than usual. Analytics have also
Vertica software is also able to manipulate the shown that Zynga’s gamers tend to buy more in-
database for social graph analysis, transposing game goods when they are offered as limited edition
all of an individual user’s interactions with other items. Zynga sells advertising, both in and around
users into a single row, and it can do this quickly. its games, but the vast majority of its revenue comes
Relational database platforms are unable to cope from its virtual goods sales.
with the massive volume of data created by all the Zynga also benefits from using Facebook as its
connections in a social graph. gaming platform. When users install a Zynga applica-
Zynga’s social graph-related data are streamed in tion, they allow Zynga access to all of their profile
real time to a dedicated Vertica cluster where the information, including their names, genders, and
graph is generated daily. Every night, the models lists of friends. Zynga then uses that information
resulting from this graph are fed back into its games to determine what types of users are most likely to
for use the next day. Zynga runs as many as 130 behave in certain ways. Zynga particularly hopes to
experiments to tweak and adjust its games each day determine which types of users are most likely to
and then observe how players react. Within minutes become “whales,” or big spenders that buy hundreds
after releasing a new feature, Zynta is able to find of dollars of virtual goods each month. Though only
out whether millions of players liked it or not. On 5 percent of Zynga’s active users contribute to cor-
the basis of this new knowledge, Zynga may make as porate revenue, that subset of users is so dedicated
many as 100 daily updates to its products. that they account for nearly all of the company’s
With this business intelligence solution, Zynga has earnings.
been able to improve the targeting of items such as Zynga’s games make heavy use of Facebook’s
gifts to effectively increase the level of interaction social features. For example, in CityVille, users must
between active players while minimizing spam to find friends to fill fictional posts at their “City Hall”
passive players. Zynga is now in a position to iden- to successfully complete the structure. All of Zynga’s
tify groups of users with similar behavior or common games have features like this, but Facebook hasn’t
paths for even more precise targeting of game-related always fully supported all of Zynga’s efforts. Zynga’s
promotions and activities. Facebook apps were formerly able to send messages
Zynga’s revenue rose from $121 million in 2009 directly to Facebook members, but they disabled the
to $1.14 billion in 2011. Clearly, Zynga’s methods are feature after complaints that it was a form of spam.
working. Traditional game-makers like Activision Still, if your friends use Zynga’s Facebook apps,
Blizzard and Electronic Arts are noting Zynga’s chances are you’ve seen advertisements encouraging
growth and success and have moved towards a you to play as well in your News Feed.
similar business model. For example, Electronic Zynga’s success has disrupted the video game
Arts launched a free Facebook version of the classic industry. Traditional video game companies begin
game The Sims. The game now has 40 million active with an idea for a game that they hope players will
monthly players and was Facebook’s fastest growing buy and enjoy, and then make the game. Zynga
app for much of 2011. begins with a game, but then studies data to deter-
Zynga’s business model is to offer free games mine how its players play, what types of players
geared towards a larger, more casual gaming are most active, and what virtual goods players buy.

MIS_13_Ch_12 global.indd 513 1/17/2013 2:30:35 PM