How to Use Data to Build and Grow Hit Games: Part 1

Data in mobile game development, argues GoPractice CEO Oleg Yakubenkov, should be treated as a forward-looking tool for making decisions, reducing uncertainty, and eliminating major product risks as early as possible, rather than as a means of reviewing past performance. Most teams, he contends, stop at topline metrics, paid marketing measurement, and A/B tests, which is insufficient for reaching the top of the grossing charts. The first of a two-part series, it offers five pieces of advice drawn from his experience on free-to-play titles such as King of Thieves, C.A.T.S., and Cut the Rope 2.

The first lesson is to test marketability before development. A 2014 ice-field combat game took four months to reach test launch and showed strong Day 1 retention above 45% but a Facebook CPI of $15 in Australia, a problem a fake app store page test could have exposed beforehand. For the follow-up, C.A.T.S., the team tested multiple art styles first, found one performing several times better than the runner-up, secured below-average acquisition costs, and reached over 100 million downloads. The second lesson is to find a niche through market analysis. By studying growth, competition strength, how leaders gained position, and where a new entrant could outperform, the author built Epic Split Truck Simulator over a weekend, reaching over 2 million downloads, 99% organic.

The third lesson identifies early signs of success or failure: playtesters who continue beyond a required 30-minute session, and internal staff who play the game voluntarily. These are signals, not definitive rules. The fourth lesson is that understanding comes from segmentation plus analysis. King of Thieves launched with 26% Day 1 and 9% Day 7 retention; comparing event sequences of retained and churned users revealed that 60% of new users never tried PvP. Forcing PvP in onboarding lifted Day 1 to 41% and Day 7 to 20%.

The fifth lesson advocates aggressive early experimentation while user bases are small. Multiplying timers fivefold and raising IAP prices for half of users dropped the store rating to two stars but raised LTV 2.5 times. Through soft-launch experiments, LTV improved more than 40-fold by global launch. The piece closes by urging constant, early experimentation and incremental gains.