Understanding difference between Targeting and Audience in AB Test
Edit me

Targeting vs Audience in A/B Testing

In A/B testing, Targeting and Audience are closely related concepts but serve different purposes. Understanding the distinction between them is critical for designing accurate experiments, avoiding data pollution, and ensuring reliable insights.

1. What is Targeting?

Targeting defines where and when an A/B test is allowed to run.
It determines whether a user is eligible to see the experiment at all.

Targeting is evaluated before a user is exposed to any variation.

Common Targeting Criteria

  • Page or URL rules (e.g., /product/*)
  • Device type (desktop, mobile, tablet)
  • Browser or OS
  • Geographic location
  • Query parameters
  • Logged-in vs logged-out state
  • Consent or privacy flags

Example

Run the test only on the Product Listing Page (PLP) for mobile users in India.

If a user does not match the targeting rules, the test never activates for them.

Key Purpose of Targeting

  • Prevents test code from running on irrelevant pages
  • Improves performance by limiting execution
  • Ensures technical correctness
  • Avoids accidental exposure

2. What is Audience?

Audience defines who should be included in the experiment analysis and variation allocation once the test is running.

Audience is evaluated after targeting conditions are met.

Common Audience Conditions

  • New vs returning users
  • User behavior (e.g., visited PDP, added to cart)
  • Traffic source (organic, paid, email)
  • User attributes (member vs non-member)
  • Custom events or data layer values

Example

From all targeted users, include only those who searched for a product.

The test may load for everyone who matches targeting, but only users in the audience are bucketed into control or variant.

Key Purpose of Audience

  • Controls experiment exposure
  • Refines who is compared in results
  • Enables behavioral segmentation
  • Improves statistical relevance

3. Key Differences at a Glance

Aspect Targeting Audience
Evaluation Before test runs After test is eligible to run
Scope Page / environment level User level
Purpose Decide if test should load Decide who is tested
Impact Performance & correctness Data quality & insights
Example Mobile users on PDP Users who added item to cart

4. Practical Example

Scenario: Testing a new CTA on PDP

  • Targeting
    • URL contains /product
    • Device = Mobile
    • Consent granted
  • Audience
    • User has scrolled 50%
    • User is not already a member

In this setup:

  • The test script loads only on mobile PDPs (Targeting)
  • Only qualified users see variations and count toward results (Audience)

5. Common Mistakes to Avoid

  • ❌ Using audience conditions for page-level control
  • ❌ Overloading targeting with behavioral logic
  • ❌ Changing audience rules mid-test
  • ❌ Assuming targeted users are automatically part of the audience

6. Summary

  • Targeting controls where and when a test runs
  • Audience controls who is included in the experiment
  • Targeting ensures technical safety
  • Audience ensures analytical accuracy

Both must work together for a reliable and meaningful A/B test.

Rule of thumb:

Target the page. Audience the user.

Tags: ab-test