Use segments to focus your audience

You can define audience segments and use them in your launches and experiments. A segment is a portion of your audience that shares one or more characteristics. Examples could be Chrome browser users, users in Europe, or Firefox browser users in Europe who also fit other criteria that your application collects, such as age.

Using a segment in an experiment limits that experiment to evaluating only the users who match the segment criteria. When you use one or more segments in a launch, you can define different traffic splits for the different audience segments.

Segment rule pattern syntax

To create a segment, define a segment rule pattern. Specify the attributes that you want to use to evaluate whether a user session will be in the segment. The pattern that you create is compared to the value of evaluationContext that Evidently finds in a user session. For more information, see Using EvaluateFeature.

To create an segment rule pattern, specify the fields that you want the pattern to match. You can also use logic in your pattern, such as And, Or, Not and Exists.

For an evaluationContext to match a pattern, the evaluationContext must match all parts of the rule pattern. Evidently ignores the fields in the evaluationContext that aren't included in the rule pattern.

The values that rule patterns match follow JSON rules. You can include strings enclosed in quotation marks ("), numbers, and the keywords true, false, and null.

For strings, Evidently uses exact character-by-character matching without case-folding or any other string normalization. Therefore, rule matches are case-sensitive. For example, if your evaluationContext includes a browser attribute but your rule pattern checks for Browser, it will not match.

For numbers, Evidently uses string representation. For example, 300, 300.0, and 3.0e2 are not considered equal.

When you write rule patterns to match evaluationContext, you can use the TestSegmentPattern API or the test-segment-pattern CLI command to test that your pattern matches the correct JSON. For more information, see TestSegmentPattern.

The following summary shows all the comparison operators that are available in Evidently segment patterns.

{
    "UserID": [ null ]
}

{
    "LastName": [""]
}

{
    "Browser": [ "Chrome" ]
}

{
    "Country": [ "France" ], "Device": ["Mobile"]
}

{
    "Browser": ["Chrome", "Firefox"]
}

{
 "$or": [
   {"Browser": ["Safari"]},
   {"Device": ["Tablet"}]
 ]
}

{
    "Browser": [ { "anything-but": [ "Safari" ] } ]
}

{
    "Price": [ { "numeric": [ "=", 100 ] } ]
}

{
    "Price": [ { "numeric": [ ">", 10, "<=", 20 ] } ]
}

{
    "Age": [ { "exists": true } ]
} 

{
    "Age": [ { "exists": false } ]
}

{
    "Region": [ {"prefix": "us-" } ]
}

{
    "Region": [ {"suffix": "West" } ]
}

Segment rule examples

All of the following examples assume that you are passing values for evaluationContext with the same field labels and values that you are using in your rule patterns.

The following example matches if Browser is Chrome or Firefox and Location is US-West.

{
  "Browser": ["Chrome", "Firefox"],
  "Location": ["US-West"]
}

The following example matches if Browser is any browser except Chrome, the Location starts with US, and an Age field exists.

{
  "Browser": [ {"anything-but": ["Chrome"]}],
  "Location": [{"prefix": "US"}],
  "Age": [{"exists": true}]
}

The following example matches if the Location is Japan and either Browser is Safari or Device is Tablet.

{
 "Location": ["Japan"],
 "$or": [
   {"Browser": ["Safari"]},
   {"Device": ["Tablet"]}
 ]
}

Create a segment

After you create a segment, you can use it in any launch or experiment in any project.