Audience filters

Filters define who qualifies for an audience. Combine customer attributes, behaviors, related records, traits, and other conditions to express the audience you want.

Add and combine filters

Add filters for the customer data or behavior that determines whether someone belongs in the audience.

You can build the conditions directly, start from an audience template, or ask the agent to draft or refine the logic.

Combine conditions with AND and OR

Use AND when all conditions must match. Use OR when any condition can match.

For example, to include users who either made a purchase or added an item to their cart, but only if they belong to the Gold tier:

Nest conditions within a related model

Conditions on the same related record can have their own AND/OR logic. For example:

has a Purchase where (amount > 100 AND category = "Electronics") OR (amount > 50 AND category = "Accessories")

Open the ... menu on the related-model filter and select Add "OR" group to add another condition group inside it.

Filter on related records

Use related and event models when the audience depends on data that isn't stored directly on the parent model.

For example, with a Users → Orders relationship, you can build an audience of users who placed an order worth more than $100. Customer Studio evaluates the related records, then includes the parent records that meet your conditions.

Select Add filter, then Relations to filter on a related model.

Set how many related records must match, then add conditions on the related model's fields.

Filter on events

Event filters target customers by what they did and when. Add an event condition, then refine it:

• Frequency — require the event at least, at most, or exactly a number of times (for example, purchased at least one time).
• Time window — limit to a period such as the last 30 days, using a rolling or calendar window.
• Where — filter on the event's own properties, such as category equals "Toy".
• Sequential events — add Then did perform or Then did not perform to require or exclude a later event within a time window, such as viewed a product and then did not purchase.

Filter by audience or journey membership

Reuse an existing audience or a journey as a condition.

Audience membership — filter by whether customers Are included in or Are not included in another audience. Use this to build on top of an existing segment or to exclude one.

Journey participation — target customers by where they are in a journey:

Compare columns and traits

Instead of comparing a field with a fixed value, you can compare two columns or traits.

For example, lifetime_value > avg_order_value includes customers whose lifetime value is greater than their own average order value, rather than comparing every customer against the same threshold.

When configuring the condition, select a value, property, or trait as the comparison, then choose the field you want to compare it with.

Build a trait inside the audience

When you need a calculated value for a single audience, add a trait condition directly in the builder instead of creating a reusable trait first. Choose a calculation method such as Average or Count, the column to calculate, and the event or related model to calculate across, then compare the result.

Select Add filter, then Traits to create a custom trait or start from a trait template.

For logic you'll reuse across audiences, create a saved trait instead.

Filter across multiple related models

For data that spans multiple relationships, you can nest related or event filters. For example, with a User → Purchases → Items schema, you can find users who made a purchase in the last 30 days that included an item in the Toy category.

Each level applies to records from that model:

• The outer condition finds qualifying purchases for each user.
• The inner condition finds qualifying items within those purchases.

A user qualifies when at least one record satisfies the complete nested path.

You can also apply counts at multiple levels. For example, find users with at least three purchases where each qualifying purchase contains at least two matching items.

Nested filtering supports up to three levels from the parent model.

Segment across people, households, and accounts

If your schema includes multiple entity types, such as people, households, or accounts, you can use relationships between them when building an audience.

For example, you could build a People audience that includes:

• people in households with at least one active subscription
• people associated with accounts that recently purchased or renewed

See Schema for how entity labels affect the audience-building experience.

Filter nested JSON

If a column contains a JSON object or array, use dot-path notation to filter on values inside it. For example:

• booking.product.geo.city targets a nested object property.
• booking.0 targets the first item in an array.

Given:

{
  "inventory": {
    "on_order": 50,
    "stock": 750,
    "warehouse": "WHB"
  },
  "dimensions": [9, 8, 7]
}

you could filter on inventory.stock, inventory.warehouse, or an item in dimensions.

JSON columns initially use the unknown type. Set the nested value's type to String, Number, or Boolean to use the operators that type supports. For example, a numeric value can use greater than, while a string can use contains. JSON filtering is supported on parent, event, and related models.

Add a list of values

For filters that accept multiple text or numeric values, you can drag a .txt or .csv file into the value field instead of entering each value manually.

This is useful for long lists such as customer IDs, organization IDs, or email addresses. The file should contain comma-separated values.

Supported operators:

Filter operator reference

The available operators depend on the field's data type.

String operators

Use these operators for text values such as product names, locations, email domains, or status fields.

contains matches the value anywhere within the field. equals matches the complete value.

Number operators

Number filters can compare a field against either:

• a fixed value, such as purchase_amount > 100
• a percentile, such as customers in the top 10% of lifetime value

To use percentiles, value suggestions must be turned off for that column in the schema configuration.

Timestamp operators

Timestamp filters support fixed dates and relative time windows. For relative periods, choose between:

• Rolling periods — calculated backward from the current time. For example, within previous 2 months.
• Calendar periods — completed calendar units. For example, if today is June 3, within previous 2 calendar months covers April 1 through May 31.

Use anniversary for recurring dates such as birthdays, renewals, or membership milestones.

Configure null behavior

Workspace admins can control whether null values match negative operators such as does not equal. For example, consider Country does not equal USA. Depending on the workspace setting:

• Treat null values like any other value — records where Country = null qualify.
• Use special behavior for null values — records where Country = null don't qualify.

Configure this behavior in Workspace Settings.

Troubleshoot filters

A filter returns no results

First check:

• that you're filtering on the expected model
• that the relationship connects the records you expect
• that the field contains the values you're filtering for
• how your workspace handles null values

For related or event data, also verify the schema relationship. To spot-check the underlying data, inspect individual members in Profiles.

Filter value suggestions are missing

Column suggestions may be disabled or out of date.

Confirm that value suggestions are enabled. If they're enabled but values are missing, refresh the suggestions from the filter.

A filter references a missing column

If a warehouse column was renamed or removed, an existing audience can continue to reference the old field, and the preview fails with an error that the column can't be resolved or no longer exists.

1. Open the model and confirm the column exists.
2. Refresh or re-save the model so Customer Studio has the current columns.
3. Update any audience conditions that reference the missing field.
4. Preview the audience again.

If the failure surfaces as a warehouse SQL error, see Resolve SQL compilation errors.

Next steps

• Build an audience to review the full audience-building workflow.
• Audience builder prompting guide for prompting patterns and guidance on using the agent.
• Create traits for reusable calculated values you want to use across audiences.