A well-known online game producer that launched the line of browser games faced a need to maximize monetization of one of his games through a thorough analysis of gamers’ data (purchases, activities, etc.) as well as need to make internal game security stronger due to up-to-the-minute detection of fraud and cheating activities. Despite having an analytics team, a vast majority of data was processed manually and data literally was not bringing any added or business value. So, the key intention was to build two analytical modules that supposed to work separately, but in the same environment and ecosystem; yet another intention lay in extensive advisory services provided by Unicsoft about the auditing current one and later, in setting up suitable and flexible DWH/ETL solution.
As a result of Business Analysis stage, Unicsoft identified following deliverables as required to fulfil set business goals:
Gaming analytics: given a historical data for entire game users and after its preprocessing (cleansing, enrichment, etc.) a cluster of active users has been defined. Then, key features of gaming behavior and in-game purchase occurrences were carefully analysis with the purpose to tune up gaming AI: main thing was to keep balance between game difficulty for keeping gamer involved and adding several “hard spots” which might be passed easily if particular in-game purchase made.
Anti-fraud module: analysis of suspicious activities occurred during gameplay or internal currency purchase: detecting fake cards and adding them into database; detection and consequently prediction of cheating activities in game process (multiple accounts, fake teams, etc.)
DWH/ETL auditing and improvement: regardless of having plenty of various datasets, all of them were stored in old and rather unstable SQL-type solution, so need of rapid access arose eventually; after thorough analysis of requirements, Clients’ data was successfully transferred to BigData solution that consisted of Hadoop + Hive alongside with analytics modules plugged onto.
Unicsoft set up a team comprised of senior-level data scientists, data warehouse specialists and business analysts with specialty in BigData field; further, a support team for model tuning and maintenance was set up as well.
The breakeven was reached after 12 months due to both modules: whilst almost right off the bat anti-fraud module gave a hand by dealing with cheaters, analytics module gradually helped in AI improvement and launching of particular in-game purchase sets dedicated to most promising groups of users eventually leading to significant increase in profits. Consequently, it led both of us to mutually beneficial cooperation.