How Audiences and Traffic Distribution Work in A/B Tests

When running an A/B test, it’s important to understand how audiences and traffic distribution interact. These two settings work together, but they are applied in a specific order.

1. Audience Defines Who Enters the Test

When you set an audience, you are defining the total population eligible for the experiment.

• If no audience is set → all visitors are eligible
• If an audience is set → only visitors who match those conditions (e.g. location, UTM, device) are included in the test

Visitors who do not match the audience will not be part of the experiment at all.

2. Distribution Defines How That Audience Is Split

Once a visitor qualifies for the test (based on the audience), they are randomly assigned to a variant according to your traffic distribution settings.

For example:

• Audience = “US visitors only”
• Distribution = 50/50

→ Only US visitors enter the test, and those visitors are split evenly between variants.

3. Important Clarification

Traffic distribution is applied only to users who are included in the test, not to all site visitors.

This means:

• Your distribution (e.g. 50/50, 90/10) applies within the audience
• The size of your audience determines how much traffic actually reaches each variant

4. A/B Testing vs Personalization

This behavior differs from Personalization:

• A/B Testing
  • Audience → defines who is in the experiment
  • Distribution → splits that group into variants
• Personalization
  • All users are eligible
  • Audience rules determine which variant each user sees

5. Best Practices

• Use audiences when you want to test a specific segment (e.g. paid traffic, mobile users)
• Use distribution to control how traffic is allocated within that segment
• Keep in mind that smaller audiences will result in lower traffic per variant