How data scientists have changed the way games are designed at Candy Crush developer King

Vincent Darley, VP of data analytics and BI for King, told Computing's Big Data & Analytics Summit how data scientists and analytics are key to Candy Crush player retention

Data scientists form a key part of the level design process for Candy Crush games and big data analysis has changed how developer King builds its mobile games, which have over 150 million monthly active players.

That's what Vincent Darley, VP of data analytics and BI for King Digital Entertainment, told Computing's Big Data & Analytics Summit during his keynote presentation.

Darley explained how analysing data surrounding experimental changes in level design for Candy Crush Soda Saga led to permanent changes to how the game is played.

"With Candy Crush Soda Saga - which was initially released to a handful of countries around the world in the middle of last year, then launched worldwide towards the end of last year - we spent a lot of time trying to tune the levels in the game," he told the audience gathered at The Waldorf Hilton in central London.

"Sometimes we're just trying to tweak little ways, but every now and again we think let's try a whole new experience for the first 30 levels of the game, let's see what that does," Darley continued, before explaining how King determines whether a change has been a success.

"One measure is second day retention; of the people who install the game today, how many come back tomorrow and play. And another one is 14 day retention; people who've installed the game today, how many come back two weeks later and are still playing," he said.

When King experimented with a set of more challenging beginner levels for Candy Crush Soda Saga, data scientists analysed the results.

"Second day retention went down, which was obviously not the idea. However, we let the test run because you always learn good stuff, and in fact 14 day retention was higher," said Darley.

"So, even though we, in some sense, frightened off some people from the first day, the overall experience from the bulk of the population was a better one and it's the long-term perspective that we care about at King."

So, analytics showed that the changes were worth keeping because they improved retention in the long run.

"That was quite a surprise, because we hadn't seen that before analysing the data, so that changed our mind-set about how we go about the design of the levels," said Darley, who added that now data scientists play a direct role in designing Candy Crush games.

"Now designing the levels is very much a collaborative exercise between data scientists who are putting a lot of data and perspective into the equation, and then the creative game designers on the other side. Those two sets of people come together and work very, very closely," Darley said.

"So it's just changed the perspective of those discussions in quite important ways," he concluded.

Exclusive Computing research launched at the Big Data & Analytics Summit demonstrates how when it comes to big data, the revolution in self-service analytics starts now.