Talk Title: Using Lean Analytics to Make Better Games

Effective talk: Market with a lower budget

Year of Talk: 2019

Video Link: Using Lean Analytics to Make Better Games

Lean Analytics:

• Getting enough data can be extremely expensive. So it can pile up.

• Make sure the team defines exactly what it needs to find and what data to collect.

• If there is a system that can do something better than you can develop, then buy it.

• Make sure the data actually helps the team. If the data helps determine whether players are enjoying the game or are coming back is valuable data.

• But if no one is using it, then it's not as useful.

• Whenever working on a game or analytics, make sure to remember the big picture. How can the analytics the team is currently collecting help the current game and future games?

• Social point has the following chart when deciding on an internal or external tool:

  
  

• When implementing a new feature, the team starts defining and implementing the feature. 

• The analytics team will then work on defining how they will track player data and how they will measure success.

•  At this point, the mechanics are pushed to the live servers, and all the data is tracked for later analysis. 

• All the data is passed through a data link into a table, which can be processed at a later date into understandable tables and charts.

• This new system helped free up engineers as they could work on new measurement systems for other mechanics or analyze the data for other game mechanics while the current data is collected.

• Data is even more useful at launch as it can help you smooth over problems before said launch and reduce problems.

• There are also certain events that tend to repeat throughout multiple games (I.E, login, purchase, etc). So once the team learns the best way to solve it, then they can work on other areas.

• Each studio can have differences in what matters for analytics and what they want the players to feel/do. Make sure to find that out.

Get Player data without players:

• It can be difficult (especially for smaller teams) to get players to test the game at the desired rate. But it's necessary, as it helps ensure the design works and there aren't any obvious problems with the game and mechanics.

• A solution is to use an AI/computer that can play the game. Doing so lets you have the game run on a computer overnight or while people work on other computers.

• Having team members or testers play through the game also gives the team a group of real people who need to think through problems and see how people might react to changes.

• When testing with people, it's important to understand what the team already knows. If it's already understood that a lot of people will do a certain action, then it can be turned into a percentage and tested with a smaller testing group. This helped reduce testing from around a month to 1-2 weeks.

• Also, try to understand how likely certain types of players might appear. Some testing would get affected if there is an abnormally high amount of spenders compared to other groups. This greatly affects the values being tracked.

• Determine how likely certain types of players might play the game. It can help drive certain features to be finished first so that the larger groups enjoy it sooner.