GAMEPERF07-BP03 Regularly monitor matchmaking performance

One of the most noticeable ways to optimize the performance of a game for players is to reduce the time that they must wait before they can enter a game session. Long wait times can cause players to lose interest and lead to attrition, so it is important to consider this when designing your matchmaking solution.

Level of risk exposed if this best practice is not established: High

Implementation guidance

When you are designing your matchmaking configuration for your game, create rules that determine the conditions that are applied to form a match. You should consider the impact that these rules will have on the performance of the system, particularly the wait times for players. Before deploying changes to your matchmaking implementation, such as the addition of new matchmaking conditions or filters, test this beforehand or consider releasing this change gradually to a small sample population of players as a canary or A/B test to gather performance metrics before introducing this change to the entire player population.

Configure your matchmaking service to generate detailed logs to understand the conditions or rules that were applied to each matchmaking request. This assists in the review and to adjust matchmaking implementation as necessary.

For example, Amazon GameLift FlexMatch provides a fully managed matchmaking service which can be used as a standalone service with your own game server hosting or used with game servers hosted on Amazon GameLift. FlexMatch can generate event notifications to Amazon EventBridge, see Set up FlexMatch event notifications. Use Amazon Simple Notification Service (Amazon SNS) to receive matchmaking data in JSON format, allowing you to automatically process and store this information for analysis to improve matchmaking performance.

Set up metrics to track how long your matchmaking service takes to find a suitable game session for players. Review matchmaking duration metrics regularly and correlate these times with player behavior and community sentiment. Use this data to develop suitable thresholds for matchmaking timeouts that can be included in your matchmaking rule configuration.

For example, Amazon GameLift FlexMatch provides support for defining matchmaking request timeouts as well as creating matchmaking rules that can allow requirements to relax over time. This feature allows you to create matchmaking that can adapt to make it straightforward to create matches and place players into game sessions when matches are difficult to find.