3 Ways Predictive Analytics Will Change Mobile Games: Part I

Predictive analytics, already transformative in industries such as e-commerce, streaming, and telecom, is poised to reshape the mobile games industry, according to GameRefinery CEO Markus Råmark in the first installment of a three-part blog series published in February 2015. The central argument is that data-driven prediction will make game development resources far more efficient by forecasting a game's commercial success before significant investment or soft launch.

The article frames the problem through the crowded App Store market, which held over 300,000 games at the time, most with little chance of success. Rovio reportedly built 51 games before achieving a hit with Angry Birds, and the best free-to-play publishers typically devote three to six months to soft launch testing. As a cross-industry parallel, the piece cites AT&T, whose data science team analyzed hundreds of television ads, coding features such as humor type, character interaction, and storyline, to build a model forecasting ad performance. The resulting campaign was described as the most successful in the company's history, generating over $50 million in sales.

Applying the same logic to games, the author contends that concept-stage titles can be evaluated and optimized by feature set. Without large user acquisition budgets or cross-promotion, a game's baseline chance of reaching the top 200 grossing list is estimated at roughly 0.01% to 1%. Drawing on GameRefinery's own database, games whose features were optimized through its predictive analytics engine reportedly have a 10% to 40% probability of reaching that list. The implication is that studios could achieve with two or three games what would otherwise require fifty or more.

The claims come from a vendor with a commercial interest, and the methodology behind the probability figures is not detailed. The post covers the global mobile free-to-play market and previews two further parts of the series.