Like it or not, gamification is on the rise. Since 2010, big consumer brands such as NBC, Walgreens and Southwest Airlines have all launched major projects that center around gaming. Why? Because games make previously dry subjects, such as corporate training, more fun and engaging.
But the gamification of the workplace comes with a few challenges, primarily in the areas of fraud and security. It’s possible to take a look at the casual and online gaming communities, where the concept has its roots, to identify potential pitfalls and solutions.
For example, how big is the problem of cheating in online games? Valve Corporation’s game platform, Steam, developed an anti-cheat solution in 2006 after it detected 10,000 cheating attempts in a single week. As of 2012, it had terminated more than 1.5 million accounts within the 60 games running on Steam.
My company, the Alacer Group, recently had the opportunity to work with one of the world’s largest casual online gaming companies to address revenue leakage experienced through fraudulent game play. The game itself has a simple premise: through social networking with friends, players can amass wealth and gain desired status. Additionally, the user may purchase in-game currency that enhances game play or adds new dimensions. Since the basic game is free, currency plays a key role in the company’s ability to generate revenue.
But when savvy social game players figure out a hack to advance without purchasing currency, there are two big repercussions. Not only does the game developer lose revenue, it loses additional players who get frustrated when they don’t advance in the game as quickly as their cheating counterparts. Normally, network security flaws involving game play are identified and solved on the back end. Often using thousands of servers at high capital and operational expense, IT security personnel will play back every transaction and analyze it in order to determine who the cheaters are and how they are manipulating the game.
This is not only time-consuming, it’s expensive. Here’s the solution we proposed: almost any online company, particularly a game developer, has vast amounts of unstructured data at its fingertips. What if that data could be manipulated to identify and stop fraudulent game activities in real time? Game analytics has emerged as one of the main resources for ensuring game quality, understanding consumer behavior, and maximizing the player experience, similar to how the film and television industries are using big data to cater to viewers.
We used this concept to examine and model game play and to determine player navigation. It was then possible to design algorithms that could define the average player and predict the average way he or she would progress in the game. We could then define the threshold range for what would be acceptable play; anyone that fell outside of the threshold range would be flagged as a potentially fraudulent player. This allowed the game developer to immediately freeze accounts for those advancing too quickly — thereby plugging the revenue hole.
The average player profile was determined by examining data such as: friends on the social network that the player interacts with; the level and rapidity of gaming achievements; and most utilized game elements. From these data points and others, three specific player types emerged:
- “one-try wonders” who only try the game once
- “early defectors” who do not utilize the game for its expected lifetime
- “try hards” who will continue to play until they obtain a goal
Of these, the “try hards” are often the most valuable players in online games, as they stay on the site the longest, are the most likely to purchase game currency and often invite friends to join. However, this is also the group that is the most frustrated by cheaters. As one user complained to developers of the online gaming platform Kabam, “You are making good players either leave the game or resort to the same cheating to try and rid the game of the ones in question.”
Unstructured data can also be used to resolve other gamer and developer frustrations. Take a cheater who establishes fake accounts, where imaginary players lose at a game in order to artificially boost the cheater’s standings. We can use deep analytics to reveal and eliminate these fake accounts based on their win/loss ratios.
Perhaps more importantly, data can also pinpoint game “whales,” or frequent players who spend too much too quickly to maximize their advancement in the game. There is a strong probability that these whales are, in fact, using stolen credit cards. It may seem odd to non-game players that someone would risk criminal charges to purchase virtual items, but it does happen — a lot. Last year, a woman in Tennessee used a stolen credit card to purchase $4500 in virtual buildings, crops, and animals in the Facebook game, Farmville.
Specifically, our gaming client used the extracted data to close security holes. Every online game has vulnerabilities; the trick is to find them quickly before they can be massively exploited. The older server-based methodology of searching for fraud can take months; instead, we developed the algorithm for identifying a potentially fraudulent player and inserted into the network stream within weeks.
It should be pointed out that not every game developer chooses to ban a fraudulent player when she is identified. In fact, one company, Rockstar Games, chose to create a second version of the online game Max Payne 3 specifically for fraudulent players; anyone found cheating in the original game was quarantined to the cheaters version.
The experiences outlined within the casual gaming industry can certainly apply to other consumer and enterprise products who use gaming tactics in their companies and products. With the explosion of applications and online sites using elements of gamification, ranging from corporate training modules to calendars tracking fitness goals, developers will face the same security issues prevalent within the gaming industry: identity theft, rogue servers, cheating players and more. Unstructured data can be one of the tools used to limit fraudulent activity.
