Audience Setup

Composable Audiences is mParticle’s zero-copy solution for segmentation. Instead of copying data into the mParticle CDP, mParticle queries your cloud data warehouse (Snowflake, BigQuery, Redshift, or Databricks) directly to build and refresh audiences.

This page explains how to create, configure, preview, activate, and manage Composable Audiences. You’ll learn how to define membership criteria, select a refresh cadence, preview membership, and activate outputs.

For a high-level introduction to Composable Audiences, see the Overview page. For common questions, see the FAQ.

Before you begin

Before creating a Composable Audience, make sure you’ve completed the warehouse setup process.

• A data engineer or database administrator should set up your cloud data warehouse connection and create at least one User Data Model.
• (Optional) They may also create and link Event Data Models or Generic Data Models to have additional data available to enrich your audiences.

Once this setup is complete, a marketer or analyst can use those models to build and activate audiences.

Step 1: Create a new audience

1. Navigate to Segmentation > Audiences in the left-hand navigation.
2. Click + New Audience.

3. Under Start from scratch, open the Select Audience Source dropdown.

4. Scroll past In-Platform Audience until you see the Composable Audience section, and select the warehouse connection and user data model you want to use.

5. Click Next.

Step 2: Enter a name and refresh frequency

1. Enter a descriptive name for your audience.

   • If you check Use Audience name in external connections, the name you enter will also be used in downstream integrations.

2. Enter a refresh frequency and start date. This determines how often mParticle runs your data model’s SQL query to add new qualifying users and remove users who no longer meet the criteria. The start date determines when mParticle will begin processing audience memberships.

   • Composable Audiences can be refreshed Hourly, Daily, Weekly, Monthly, Once, or On Demand.
   • Composable Audiences cannot be refreshed in real time. Audience membership is updated only on the schedule you define.

Step 3: Enter audience criteria and conditions

Under Define Your Audience, describe who should be included in your audience based on fields from your data model (and any related models).

• Criteria specify the type of data you want to evaluate, such as a user attribute (loyalty_status), an event (purchase), or a generic attribute (subscription_plan).
• Conditions (or matching rules) are the rules that criteria must meet for a user to be included. For example, loyalty_status equals “Gold”, or purchases_last_30_days greater than 3.

By combining criteria and conditions, you can precisely define which users belong to your audience.

As you build, click Get Size at any time to preview how many users currently match your criteria. This helps validate your selections and logic.

Example: At-Risk Subscribers

Suppose you want to build an audience of at-risk subscribers. In the builder:

1. Click Select criteria and choose the attribute engagement_days_last_30.

2. Set the condition to less than 3.

3. Click + Add Criteria and choose churn_risk_score.

4. Set the condition to greater than 0.7.
5. Keep the operator as AND so both must be true.

Click Get Size to preview how many users qualify. Once satisfied, save and activate the audience for a retention campaign.

Data types and matching rules

Each field you select has a data type, which determines the conditions you can apply. For example, a string field supports conditions like equals or contains, while a number field supports comparisons like greater than or less than.

Supported data types:

• String
• Number
• Boolean
• Date / Timestamp

Matching rules by data type

When building Composable Audiences, the criteria available to you are determined by the columns defined in your data model. Each column has a data type, which controls the matching rules you can apply.

Strings

String attributes support the following matching rules:

• Contains: Matches when the specified substring is present.
  Example: "blue" matches both "blue" and "blue shirt".
• Matches: Matches when the string matches a specified pattern.
  Example: The pattern "bl?e" matches "blue" and "blee", while "b*e" matches "blue", "brave", and "barbecue".
• Equals / Does not equal: Matches when the entire field exactly equals or does not equal the specified value.
  Example: "blue" equals "blue", but does not equal "blue shirt".
• Greater than / Less than / Greater than or equal / Less than or equal: Compares the string value in lexicographic order (alphabetical sorting).
  Example: "apple" is less than "banana".
• Starts with / Does not start with: Matches when the string begins (or does not begin) with the specified value.
  Example: "blueberry" starts with "blue".
• Ends with / Does not end with: Matches when the string ends (or does not end) with the specified value.
  Example: "lightblue" ends with "blue".
• Exists / Does not exist: Matches based on whether the attribute is present in the dataset.
• Is null / Is not null: Matches based on whether the attribute has a stored value.

Numbers

Number attributes support equality and range comparisons:

• Equals / Does not equal: Matches when the number exactly equals or does not equal the specified value.
  Example: 10 equals 10 returns true, while 10 does not equal 5 also returns true.
• Greater than / Less than / Greater than or equal / Less than or equal: Matches when the number is higher, lower, or equal to the specified value.
  Example: orders_count greater than or equal 5 finds users with five or more orders.
• Exists / Does not exist: Matches based on whether the numeric attribute is present.
• Is null / Is not null: Matches based on whether the numeric attribute has a stored value.

Booleans

Boolean attributes support simple truth-value checks:

• Is true: Matches when the field is true.
  Example: opted_in_to_marketing is true includes users who have opted into marketing communications.
• Is false: Matches when the field is false.
  Example: opted_in_to_marketing is false excludes users who have opted in.

Timestamps

Timestamp attributes support calendar-based comparisons:

• Is on / Is not on: Matches records that fall exactly on or not on a given date.
  Example: signup_date is on 2024-01-01 finds users who signed up that day.
• After / Before: Matches records later or earlier than the specified date.
  Example: last_purchase_date before 2024-01-01 finds users whose last purchase was in 2023.
• After or on / Before or on: Matches records on or after, or on or before, the specified date.
  Example: renewal_date before or on 2024-12-31 includes users whose subscriptions expire by the end of 2024.
• Exists / Does not exist: Matches based on whether a timestamp attribute is present.
• Is null / Is not null: Matches based on whether the timestamp field has a stored value.

Groups and logical operators

Groups let you bundle multiple criteria together and control how they’re evaluated with AND or OR logic. They’re useful for building more precise segments. For example, grouping subscription tiers (Premium OR Business) and then combining that group with purchase activity (AND made a purchase in the last 30 days). Using groups can help you to organize complex audience definitions.

• Use AND when all conditions must be true.
• Use OR when only one condition needs to be true.
• Drag and drop criteria to reorder or nest them.
• Click + Add Criteria to add additional conditions or groups.

Step 4: Save and activate your audience

When you’re finished, click Save Audience.

You can edit your audience definition or refresh cadence at any time.

Once saved, you can:

Activate the audience by clicking the three-dot action menu and clicking Activate so membership changes are updated according to your refresh cadence.

Connect the audience to a downstream integration by clicking the + button under Connect on the Audiences landing page and selecting one of your configured outputs.

The process for connecting and activating composable audiences is the same as for Real-Time Audiences. For more information about connecting your new composable audience to downstream integrations, see Connect an Audience.

Key differences between Composable and Real-Time Audiences

• Audience sizes shown in the UI are actual sizes, not estimates.

• There are also some differences between the Audience Builders available for each type of audience:

  • The Audience Builder for Composable Audiences has clearer logical groupings and supports drag-and-drop to rearrange criteria. These features aren’t available in the Real-Time Audiences Builder.
  • The Real-Time Audience Builder currently supports features such as Audience Insights, Audience Groups, and A/B Tests. These features are not yet available in the Composable Audience Builder, but they are planned for a future release.
• Composable Audiences are clearly labeled as such on the Audiences landing page.