Skip to Content
GuidesSegmentsCreate a segment

Create a segment

Go to People → Segments → Create to open the visual segment builder.

Name your segment

Give it a descriptive name (e.g., “High-value Mumbai customers” or “Cart abandoners — last 7 days”). The name appears everywhere segments are referenced — campaigns, journeys, analytics.

Add conditions

Click Add condition to add a rule. Each condition has three parts:

  1. Field — what to evaluate

    • Contact property: name, email, city, plan, any custom field
    • Event: has done / has not done a specific event
    • Event column: event.location_id, event.workspace_id — top-level event columns evaluated against every row in the event stream (distinct from event-property filters which probe the JSONB payload)
    • Engagement: opened email, clicked WhatsApp, etc.
    • Segment membership: is in / is not in another segment

  2. Operator — how to compare

    • Text: equals, not equals, contains, starts with, is set, is not set
    • Number: equals, greater than, less than, between
    • Date: within last N days, before, after, on
    • Event: has done (with optional count and time window)

  3. Value — the comparison target

Combine with logic

Multiple conditions combine with AND or OR:

  • AND — contact must match ALL conditions
  • OR — contact must match ANY condition

For complex logic, create condition groups: conditions within a group use AND; groups combine with OR (or vice versa).

Example:

(city = "Mumbai" AND plan = "pro")
OR
(total_orders > 10 AND last_order within 30 days)

Preview

The builder shows a live count of contacts matching your rules. Click Preview to see a sample of 10 matching contacts with their properties.

Save

Hit Save segment. The segment starts evaluating immediately. Contacts matching the rules are added; those who stop matching are removed — all in real-time.

AI segment generation

Don’t want to build rules manually? Click Generate with AI and describe your audience:

“Customers who spent more than ₹5,000 in the last quarter but haven’t purchased this month”

The AI translates this into conditions, shows them to you for review, and lets you edit before saving.

Natural-language outlet names

Multi-outlet chains can reference outlets by name in the prompt — the parser resolves the name to the canonical location_id:

“Lapsed Juhu customers who used to order weekly”

“Members of the MG Road outlet who haven’t bought in 30 days”

Phrases like “at Juhu”, “per outlet”, or “across brands” map to the outlet and brand_workspace concepts in the parser’s concept registry, which lower into event.location_id and event.workspace_id conditions. Ambiguous outlet names (e.g. two stores both called “Central”) surface clarification chips before the segment is built.

Tips

  • Start broad, then narrow. A segment with too many conditions often matches zero contacts.
  • Use the preview to verify before saving — catch mistakes early.
  • Avoid deeply nested logic (5+ conditions). Split into multiple segments and use “is in segment” as a condition for composition.

What’s next

Segments overviewSegment detail & analytics