Skip to end of metadata
Go to start of metadata

While A/B testing and control functionality can be configured for each campaign, sometimes you want to ensure a visitor falls into a specific group across multiple campaigns, or you want to split campaign traffic into halves, thirds, quarters, fifths or some other interval. You can use A/B test segments to accomplish this.

This Article Explains

This article details how to create an A/B test segment, shares some uses of this type of segment, and gives guidelines for creating more popular intervals to create an A/B test segment, shares some uses of this type of segment, and gives guidelines for creating more popular intervals.

Sections in this Article

Uses of A/B Test Segment

Although you can A/B test any campaign in Evergage, using A/B Test Segments gives you additional flexibility and control over how traffic is distributed and which users see which experiences. Campaign scenarios that lend themselves well to A/B Test Segments typically involve more than one campaign:

  • Your strategy includes an initial campaign followed by subsequent campaigns at a specific interval. For example, suppose you are testing whether subsequent recommendations campaigns have better lift over control. You want the same group of people to receive the campaign experience with recommendations versus the control experience which does not have recommendations. Creating a control group segment that always receives the control experience can help you accomplish this.
  • Three campaigns are part of your test - each one with a control experience. You want to make sure the same group always gets the control experience. Creating two groups - one for "Global Control" and one for "Global Experience" can help accomplish that. You can also use Segment Compare to see how the two segments perform against each other for all campaigns.


Create an A/B Test Segment

  1. Create a segment
  2. Add a Segment Name
  3. Select Campaigns as the segment Category
  4. Select A/B Test Segment as the segment Rule
  5. Enter the Total population range (see Common A/B Test Intervals below for examples)

    The upper bound on segment is not inclusive. This means that a segment looking for users from 0-25% only includes those users that are grouped between 0 and 25, not those that are exactly in the 25% group
  6. Click SAVE


Common A/B Test Intervals

Since the upper bounds on the segment rules are not inclusive, you have to adjust the structure your segments to account for that. So if you configure a segment that is looking for users between 0% and 5%, the segment would include users with values of 0, 1, 2, 3, and 4. The next segment of users between 5% and 10% would include users with a value of 5, 6, 7, 8, and 9. Each segment has exactly 5% of the users. 

50-50 Split

GroupConfiguration
First 500-50
Second 5050-100

Thirds

GroupConfiguration
First third0-34
Second third34-67
Third third67-100

Fourths

GroupConfiguration
First fourth0-25
Second fourth25-50
Third fourth50-75
Fourth fourth75-100



Fifths

GroupConfiguration
First fifth0-20
Second fifth20-40
Third fifth40-60
Fourth fifth60-80
Fifth fifth80-100