![](\"/images/guides/composable-audiences/data-models-overview.png\")
Data models define how mParticle queries your warehouse to build composable audiences. Each model is based on a SQL query and specifies which data—such as users, events, or attributes—is available for segmentation.
\nOnce created, data models act as building blocks for audiences. You can reuse them across multiple audience definitions and link them together to create more advanced segmentation logic.
\nData models define which data from your cloud data warehouse can be used to build audiences in mParticle. Each type of data model plays a different role in audience creation:
\nAll models are created from SQL queries and tied to a specific cloud data warehouse connection (such as Snowflake, BigQuery, Redshift, or Databricks). Together, user, event, and generic models allow you to combine user attributes from your warehouse, behaviors over time, and contextual data to build precise, flexible audiences.
\nThe Data Models page, located under Data Warehouse > Data Models, shows all models available in your workspace for Composable Audiences.
\n\nFrom this page, you can:
\n
To create a data model:
\nFollow the step-by-step instructions for:
\n\nIf you enable autocomplete for a column in a data model, mParticle temporarily caches the most common values from that column. These values are held for up to 7 days and then purged on a rolling basis. Autocomplete must be explicitly enabled for each column, and doing so makes it easier to build audiences by suggesting common values from your warehouse as you type.
\nThe table on the Data Models page includes key information for each model:
\n| Column | \nDescription | \n
|---|---|
| Model Name | \nThe name of the data model. Click to view or edit the model configuration. | \n
| Type | \nThe model type—either Users, Events, or Generic. | \n
| Connection | \nThe warehouse connection the model uses. Must be active and healthy for audiences to refresh successfully. | \n
| Primary Key | \nThe unique identifier for each user or record (e.g., user_id, event_id). | \n
| Relationships | \nNumber of relationships this model has with others. Relationships link models for joined audience logic. | \n
| Created | \nTimestamp of when the model was created. | \n
| Created By | \nThe user who created the model. | \n
To delete a model, hover over its row, click the three-dot menu, and select Delete.
\n\nCreating relationships between data models allows you to join datasets (such as linking events or subscriptions to users) so you can define more complex audience criteria. Relationships are established by specifying a shared field (key) that exists in both models, for example a user_id column.
Relationships must be created after both models exist. Only user models can stand alone. You must define at least one relationship to use event or generic models in an audience. For example:
\nuser_id to create an audience of users who completed a specific action.Without these relationships, mParticle would not know how to connect activities or attributes back to the correct user record.
\nImagine you have created two data models:
\nuser_id, email, and loyalty_tier.order_id, user_id, purchase_date, and order_value.By creating a one-to-many relationship between the user_id in your User model and the user_id in your Orders model, you make it possible to build an audience such as “all Gold loyalty tier users who have made more than 3 purchases in the last 90 days.”
![](\"/images/guides/composable-audiences/data-models-overview-add-relationship-1.png\")
user_id → user_id).Choose the cardinality of the relationship:
\nIn the example below, the Users data model has been linked to both the Orders event and the Subscriptions generic data model through the shared USER_ID column.
\n![](\"/images/guides/composable-audiences/data-models-overview-add-relationship-2.png\")
These relationships are powerful because they allow you to build audiences that combine different perspectives on your users. For example, “Gold loyalty tier users who made more than three purchases in the last 90 days” or “Premium plan subscribers whose renewal date is within 30 days but have auto-renew turned off.”
\nWithout these relationships, mParticle would not be able to connect purchase or subscription records back to individual users.
\n","frontmatter":{"title":"Data Models Overview","order":1,"seoTitle":null,"seoDescription":null},"parent":{"__typename":"File","modifiedTime":"October 12, 2026"}}},"pageContext":{"slug":"/guides/data-warehouse/composable-audiences/data-models/overview/"}},"staticQueryHashes":["2950460867"]}