In Data analysis, Free to play, Games Analytics

The advent of online games has put the games industry in the privileged position of being able to track every data event that a player undertakes.

However, the challenge lies in unlocking that big data to create actionable insights that can make environments more fun, boost player engagement, and drive revenues.

While the use of metrics dashboards in game design is now commonplace, they only really provide a window into the games performance and can’t solve complex retention issues.  This requires a combination of powerful data mining technology and skilfull analysts to dive deeper and get to the core of the problem.

Rarely are retention metrics improved through one big solution; rather, a series of separate issues are addressed which all add up to a well-balanced game.

The ability to follow train-of-thought paths of investigation is key. The journey to the right answer is indirect.  Data mining in its purest sense is a voyage of discovery, isolating individual behaviors through a series of general and then more detailed queries zeroing in on specific behaviors   And if the queries take several minutes or even hours to run, then the information flow is blocked.

So how do you crunch through a billion in-game events to identify the various issues that are impacting retention?   Let’s go from a billion to one.

1.  Identify where & why players are leaving

What causes one player to fall out of the game may not be an issue for player with a different level of competency or playing style, which is why retention issues are rarely caused by just one thing.

Therefore, solving retention issues requires a multilayered approach and the right tools and expertise to segment your playing base into clear playing styles.

For example, let’s say our retention metrics are low and funnels have isolated that mission four is a blocker.  The key question is, what’s happening within mission four that is resulting in players leaving?

Are they running out of resources before completing?  Are they failing multiple times and abandoning the game?

Once you’ve identified the cause of the problem, then you can begin to find the solution.

2.  Create personalized response to each problem

The solution to a problem will differ for each player segment, so if you want to maximise engagement, you need to ensure you have a personalized solution tailored to each player behavior.

Potential solutions for the example above could be to give a gift to certain players at the start of mission four or to reduce the difficulty if a player has failed the mission 10 times.  Specific solutions to specific behaviors.

3.  Leave the rest of the players alone

If you’ve identified that difficulty level on a certain mission is causing a lot of players to fall out of the game, simply lowering the mission difficulty across the player base will only make retention issues worse, because the engaged players’ experience will be impacted creating a new set of problems.

This is why train-of-thought analysis is so important.  Analysts explore hunches and need to get answers back quickly to continue their investigation down the right path, ruling out dead ends and exploring fruitful paths in detail.

4. A/B testing

As you crunch down through the data and understand which variants of your game experience you need to present for different player behaviors, you can start to balance the game so that everyone gets the best possible experience.

This is where A/B testing plays an invaluable role in establishing which solutions work best and what changes can deliver the best and often immediate impact on retention and monetization.

Is 50 gems the right level of gifting at the start of mission four so that players make it through the mission but future revenues are not cannibalized?  Or 100? Or 25? Try different scenarios in A/B testing to ensure that the game economy is well balanced.

5. Measure the uplift, & optimize going forward

Congratulations. You’ve successfully gone from a billion to one.

As all publishers and developers in free-to-play know only too well, it is very difficult to balance engagement and lifetime value.  Over-gifting can reduce revenues later on; harsh blockers reduce first-time payments; and insistent monetization offers destroy engagement.

To ensure the ongoing success of your game, you must continually measure the uplift which your design changes have on the game. An ongoing test-and-learn cycle is fundamental to delivering optimized playing experiences and maximizing the revenue potential of your game.

In billion to one, Part 2, we will look at specific examples of how data mining has leveraged better game experiences with real examples that create responsive player management.


This post originally appeared on Gamesbeat.

Recommended Posts

Start typing and press Enter to search