Targeting & Segments

Targeting lets you deliver different flag values to different users based on their attributes. Segments are reusable groups of targeting conditions.

Targeting Rules

A targeting rule consists of:

Component	Description
Conditions	Attribute-based filters (e.g., country equals "US")
Segment Keys	Reference to reusable segments
Value	The value to return when the rule matches
Percentage	Percentage of matching users to target (basis points: 0–10000)
Priority	Evaluation order (lower = evaluated first)
Match Type	all (AND logic) or any (OR logic)

Operators

Operator	Key	Description
Equals	eq	Exact match
Not Equals	neq	Inverse match
Contains	contains	Substring match
Starts With	startsWith	Prefix match
Ends With	endsWith	Suffix match
In	in	Value is in the list
Not In	notIn	Value is not in the list
Greater Than	gt	Numeric comparison
Greater Than or Equal	gte	Numeric comparison
Less Than	lt	Numeric comparison
Less Than or Equal	lte	Numeric comparison
Regex	regex	Regular expression match
Exists	exists	Attribute is present

Example: Target Beta Users in the US

{
  "rules": [
    {
      "priority": 1,
      "description": "Beta users in US",
      "match_type": "all",
      "conditions": [
        {"attribute": "country", "operator": "eq", "values": ["US"]},
        {"attribute": "beta", "operator": "eq", "values": ["true"]}
      ],
      "value": true,
      "percentage": 10000
    }
  ]
}

How Conditions Evaluate

• match_type: "all" — All conditions must be true (AND)
• match_type: "any" — At least one condition must be true (OR)

Conditions compare against attributes in the evaluation context:

client.boolVariation('my-flag', {
  key: 'user-123',
  attributes: {
    country: 'US',
    beta: 'true',
    plan: 'enterprise',
    age: '30',
  },
}, false);

Segments

Segments are reusable groups of conditions that can be referenced by multiple flags. Instead of duplicating "enterprise users" conditions across flags, create a segment once.

Creating a Segment

curl -X POST http://localhost:8080/v1/projects/$PROJECT_ID/segments \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "key": "enterprise-users",
    "name": "Enterprise Users",
    "description": "Users on enterprise plan",
    "match_type": "all",
    "rules": [
      {"attribute": "plan", "operator": "eq", "values": ["enterprise"]}
    ]
  }'

Using Segments in Targeting Rules

Reference segments by key in your targeting rules:

{
  "rules": [
    {
      "priority": 1,
      "segment_keys": ["enterprise-users"],
      "value": true,
      "percentage": 10000
    }
  ]
}

When a rule references segments, the evaluation engine:

1. Looks up each segment in the ruleset
2. Evaluates the segment's conditions against the user's attributes
3. If at least one segment matches, the rule's conditions are evaluated (if any)

Evaluation Order

Rules are evaluated in priority order (ascending). The first matching rule determines the result:

1. Sort rules by priority (lowest first)
2. For each rule:
   • Check segment conditions (if segment_keys present)
   • Check attribute conditions
   • If match and percentage == 10000: return rule value (TARGETED)
   • If match and percentage > 0: check user's hash bucket (ROLLOUT if in range)
   • Otherwise: continue to next rule
3. If no rule matches: check default percentage rollout
4. If nothing matches: return flag's default value (FALLTHROUGH)