Current Page

Current Page

Feb 20, 2024

A/B Test Segmentation

This guide explains traffic allocation and audience segmentation in A/B testing and split testing environments. Learn how to configure traffic distribution, user segmentation, and variant assignment for your website optimization and conversion rate optimization (CRO) experiments.

Segmentation Options

Location Targeting

Users are matched against their country of origin. A variant can be limited to one or multiple countries. The system evaluates the user's location against the configured countries list.

Language Targeting

Variants can be restricted to specific browser languages. The system checks the user's browser language settings against the configured language list. Multiple languages can be specified for a variant.

Browser Targeting

Browser targeting offers two methods:

  1. Vendor matching: Target specific browser vendors (e.g., Chrome, Firefox, Safari)

  2. User-Agent pattern matching: Use regular expressions to match the User-Agent header for more granular control

UTM Parameter Targeting

Variants can be restricted to users based on UTM parameters in their entry URL:

  • Supports all UTM parameters (utm_source, utm_medium, utm_campaign, etc.)

  • UTM parameters are sticky - they persist across the user's session

  • If a user navigates away from the entry URL, the system still uses their original UTM parameters for variant assignment

Traffic Allocation Process

  1. Segmentation Check

    • System evaluates all configured targeting rules (Location, Language, Browser, UTM)

    • Users must match ALL configured rules to be eligible for the variant

    • If any rule fails, the user is redirected to a different variant

  2. Ratio Application

    • After passing all segmentation rules, users are assigned based on the specified ratio

    • Ratio determines the percentage of eligible traffic assigned to the variant

  3. Original Variant

    • Captures all traffic that doesn't match other variant criteria

    • Acts as a fallback for users who:

      • Don't match segmentation rules

      • Match rules but fall outside the ratio

      • Have technical issues during evaluation

Traffic Flow Example

Given two variants with the following configuration:

Variant B (Test):

  • Location: US, UK

  • Language: en

  • Browser: Chrome

  • UTM: utm_source=email

  • Ratio: 30%

Variant A (Original):

  • No restrictions (catches remaining traffic)

A user will be assigned to Variant B only if:

  1. They are from the US or UK

  2. Their browser language is English

  3. They are using Chrome

  4. They arrived via an email campaign (utm_source=email)

  5. They fall within the 30% ratio allocation

Otherwise, they will be assigned to Variant A (Original).

200k free users on the house!

Jumpstart your A/B testing with 200k free users

We don't like giant flat rates! So we're giving you 200k free users each month. After that, only $1 for every 1000 unique visitors who participated in an A/B testing.

Stacey O.

Alex V.

Thomas P.

* Introductory pricing for the next 100 customers only!

200k free users on the house!

Jumpstart your
A/B testing with 200k free users

We don't like giant flat rates! So we're giving you 200k free users each month. After that, only $1 for every 1000 unique visitors who participated in an A/B testing.

* Introductory pricing for the next 100 customers only!

200k free users on the house!

Jumpstart your A/B testing with 200k free users

We don't like giant flat rates! So we're giving you 200k free users each month. After that, only $1 for every 1000 unique visitors who participated in an A/B testing.

Stacey O.

Alex V.

Thomas P.

* Introductory pricing for the next 100 customers only!

200k free users on the house!

Jumpstart your A/B testing with 200k free users

We don't like giant flat rates! So we're giving you 200k free users each month. After that, only $1 for every 1000 unique visitors who participated in an A/B testing.

Stacey O.

Alex V.

Thomas P.

* Introductory pricing for the next 100 customers only!