Matchmaker A/B testing#

Finding the right matchmaking rule that provides the best game experience is important for player engagement. A/B testing is a feature that allows you to test, evaluate and optimize matchmaking rules without tampering with game operations or revenue. You can experiment with different matchmaking rule configurations and compare match outcomes.

The A/B testing dashboard displays real-time metrics about how the different matchmaking rules are performing. This data helps you evaluate and decide which ones align with your goals. The dashboard displays retention, engagement and monetization data to give a full picture of each of the matchmaking rule variants. You can also use the dashboard real-time data to tweak the matchmaking rules to best fit your needs.

Requirements#

To use A/B testing:

1. Install the latest release of the Matchmaker SDK package com.unity.services.matchmaker 1.1.1.
2. Add the Analytics SDK package com.unity.services.analytics.

To set up A/B testing:#

1. In the Unity Cloud Dashboard, go to Matchmaker > Queues > A/B Testing. Select a pool and then A/B testing.
2. Navigate to the pool variant you want to do the A/B test on and select A/B Test with Game Overrides.
3. Select Create Override and then enter an A/B test name.
4. Set the target audience that you want to aim your A/B test to:
   • Audiences: Build custom stateful audiences using Unity Analytics tracking.
   • JEXL is not supported at the moment.
5. Choose your content type and create different matchmaking rule variations.
6. Set when the A/B test will run and how long it’ll run for. Adjust the start and end dates for your A/B test.

Limitations#

The current version of Matchmaker AB Testing does not support the following Game Overrides features. We are working to incorporate these features in our future updates:

• Contextual data attributes (JEXL) for player targeting are not supported at the moment.
• The Experiment summary on the Game Overrides Unity Cloud Dashboard does not show results from Matchmaker AB tests. However, users can view results from the graphs in the Experiment summary.
• The total players shown on the reporting page refers to the total number of players involved in the AB tests, but it is not equal to the total number of confirmed players who have been served by the Matchmaker configurations.