Create an Audience

Creating an audience begins with defining your use case, segmentation approach, and engagement strategy. To do this, consider the following questions:

• What is the business goal you are trying to accomplish?
• Which customer segments are important to achieving this business goal?
• Once you identify the relevant users, what is your strategy for engaging them?

Example Use Cases

Increase User Retention for a Mobile App

Retaining users is crucial for the sustained success of a mobile application. By identifying and re-engaging inactive users, you can boost overall engagement and reduce user drop-off.

Goal: Improve user retention by targeting users who have not opened the app in the last 7 days.
Segmentation Strategy: Identify users who have been inactive for a week but used the app actively within the previous month.
Engagement Strategy: Send personalized push notifications offering a discount or exclusive content to re-engage users.

Predictive Audience for Likely Purchasers

Leveraging predictive analytics allows businesses to anticipate user behavior and proactively engage those most likely to convert, thereby optimizing marketing efforts.

Goal: Increase conversion rates by targeting users predicted to make a purchase in the next 7 days.
Segmentation Strategy: Utilize Predictive Audiences to identify users with a high likelihood of purchasing based on machine learning models analyzing past behaviors and interactions.
Engagement Strategy: Deliver personalized email campaigns featuring product recommendations or special offers to these high-likelihood purchasers.

Optimize Engagement in a Subscription Service

Identifying subscribers at risk of cancelling enables proactive engagement strategies to maintain a stable subscriber base.

Goal: Reduce cancellations by identifying at-risk subscribers.
Segmentation Strategy: Identify users whose subscription renewal is within the next 30 days and who have decreased engagement (e.g., fewer logins or interactions).
Engagement Strategy: Offer these users tailored incentives such as a loyalty bonus or a discounted renewal rate to encourage continued subscription.

Create a New Audience

Once you have clearly defined your use case, it’s time to begin building your audience.

Navigate to the Audience Creation Page

1. From the mParticle Overview Map, select Segmentation.
   
2. On the Audiences landing page, click Create New.
   

Configure your audience group

Individual audiences are contained within folders called audience groups. The first step in creating a new audience is to configure this folder:

1. Enter a name for your folder.
2. Select your Inputs (the platforms and feeds that will supply data to define the audiences within your folder).
3. Click Create.

Define Your Audience

After saving your audience group, you’ll enter the Editor. Here is where you can add, view, edit, and connect audiences.

Follow the steps below to create your first audience within this folder:

1. Click the Add criteria box to open the Audience Builder Modal.

2. In the audience builder modal, choose the environment (Production, Development, or both) from which the audience will receive data. (Note the audience environment considerations below.)
3. Click Add Criteria to begin adding criteria for your audience.
4. Select the type of data you would like to use for your first criteria (more on criteria types below).
5. Referring to the data types and matching rules below, use the criteria editor to further target the users who fit your use case. .
6. Once you have added your first criteria, click Done.

After you have added your first criteria, a number displays that represents the estimated audience size: . This estimate is based on a sample of data. As you continue to add criteria, you will see an estimated size for both individual criteria as well as for the whole audience.

4. Continue adding more criteria using And, Or, or Exclude logic to further refine your audience.

Time delays

Time delays are configurable waiting periods between parent and child audiences. When a user qualifies for and is added to a parent audience, the delay postpones their evaluation for membership in the child audience until the configured duration has elapsed.

Use time delays for customer lifecycle sequences that depend on timing as well as eligibility, including nurture cadences, retargeting cooldowns, delayed re-engagement, or any part of the customer journey where engaging with the customer immediately would be premature or create a poor experience.

How time delays work

A time delay governs when mParticle evaluates members of a parent audience for membership in a child audience.

1. When a user qualifies for and is added to the parent audience, their countdown for the delayed child audience begins.
2. While waiting, the user isn’t a member of the child audience and isn’t forwarded to outputs connected to the child audience.
3. When the time delay completes, mParticle evaluates the user against the child audience’s criteria.
4. If the user qualifies, mParticle adds them to the child audience and forwards their membership to its connected outputs. If the user doesn’t qualify, they leave the waiting period without being added to the child audience.

Example: Post-purchase follow-up

Imagine you are a retailer and you want to contact customers 3 days after a purchase, but only if they haven’t made an additional purchase. You would create a Recent Purchasers parent audience for users with a purchase in the last 7 days and a Day 3 Follow-Up child audience for users with exactly one purchase during that period. Then, you would add a 3-day time delay for the child audience.

A customer who makes a purchase on Monday is added to Recent Purchasers, which starts their 3-day countdown for Day 3 Follow-Up. When the delay completes on Thursday, mParticle evaluates the customer against the Day 3 Follow-Up criteria. If they made an additional purchase, they don’t qualify and aren’t added to the child audience. Otherwise, mParticle adds them to Day 3 Follow-Up and forwards their membership to its connected outputs.

Add a time delay to an audience

To add a time delay to an audience:

1. In the Audience Group Editor, select the + icon between a parent audience and the child audience you want to add a time delay to.
2. Click Add Time Delay.
3. Enter the delay duration and select Hours, Days, or Weeks. There is no maximum duration for a time delay.
4. Click Add delay.

When you add a delay below a parent audience, mParticle applies it to each of that parent’s direct child audiences. The delay doesn’t propagate to deeper descendants.

Time delay precision

Configured time delays represent the minimum duration before mParticle evaluates a parent audience member for membership in a child audience. After the waiting period ends, this evaluation can add a short delay before a qualifying user is added to the child audience. This added latency varies with audience size.

Modifying a time delay

You can edit an individual child audience’s delay independently of its siblings. After you customize any child audience’s delay, the group-level edit control is locked, and you must manage each child audience’s delay individually.

If sibling audiences have different delays, and you add another child node, the new node receives the delay configured on the earliest-created sibling. Review the new node’s delay and adjust it if necessary.

Edit or delete a time delay

When you edit a delay, mParticle recalculates each user’s remaining waiting period from the time they originally entered the child audience’s waiting room. If the revised time delay is shorter than the time a user has already waited, mParticle immediately evaluates the user against the child audience’s criteria.

When you delete a delay, mParticle immediately evaluates every user in the child audience’s waiting room. Users who meet the child audience’s criteria are added to that audience.

Frequently asked questions

Does a time delay reset if a user drops out of the parent audience and later re-qualifies?

Yes. If the user later re-qualifies for and is added to the parent audience, a new countdown begins for the child audience. Previously elapsed waiting time isn’t preserved.

Can different audiences in an audience group have different delays?

Yes. Set the delay for the parent’s direct child audiences first, then edit individual child audiences. Customizing any individual child audience’s time delay locks the group-level edit control.

What happens if I change the delay duration while users are waiting?

The updated duration applies to users who are already waiting for the child audience. When the updated delay completes, mParticle evaluates each user against the child audience’s criteria. If a user has already waited longer than the updated duration, mParticle evaluates them immediately.

Can I stack delays across multiple audiences?

Yes. Each time a user is added to a parent audience along the path, the countdown begins for the next delayed child audience. The total time required to reach the final audience is the sum of the delays along the path, plus any processing time incurred at each step.

Do waiting users appear in audience exports or downstream syncs?

No. While users are waiting for a child audience, they aren’t members of that audience and aren’t forwarded to its connected outputs.

Why can’t I add a delay to an audience?

The three most common reasons an audience won’t allow you to add a time delay are:

1. The audience is the last one in the path and has no child audience below it.
2. The path contains an A/B test.
3. One of the audiences in the path is not a real-time audience.

Create an A/B Test

Audience A/B testing randomly divides an audience into mutually exclusive variations so you can compare different targeting approaches. Each variation is represented as an individual audience and can be connected to outputs independently.

For example, if you want to reengage an audience of users with low engagement, you could create this test:

• Send 40% of the audience to Messaging Platform A.
• Send 40% of the audience to Messaging Platform B.
• Keep the remaining 20% as an unmessaged control group.

You can then compare engagement across the variations and use the most successful strategy for the full audience.

Add Test Variations

1. In the Audience Group Editor, select the option to add a path beneath the audience you want to test, then select A/B Test.

2. Add and name up to five test variations, then assign the whole-number percentage of audience members that each variation should receive. Any percentage not assigned to a test variation is assigned to the control group.

3. Review the variation names and allocations, then save the test.

After you save the test, each variation appears as a separate branch in the Audience Group Editor. Audience members are assigned to one variation, including the control group, according to the percentages you configured.

Connect Test Variations

Each variation, including the control group, is created as an individual audience with its own Connect Output action. You can connect different variations to different outputs or leave the control group unconnected. You can also connect the original audience before the A/B split to an output.

For detailed connection instructions, see Connect A/B Test Variations.

Manage an A/B Test

You can edit the source audience definition without removing the test. As audience membership changes, mParticle continues to assign qualifying users according to the percentages configured for the test.

To end an A/B test:

1. In the Audience Group Editor, select the option to add a path directly above the test branches, then select Delete A/B Test.

2. Review the deletion warning, then select Delete.

Deleting an A/B test also deletes its variations’ audiences and output connections. mParticle stops forwarding data to those outputs, which may leave their downstream audiences out of date or out of sync.

Audience Refresh Frequency

The Refresh Frequency setting determines how often mParticle updates audience membership. Choosing the right refresh frequency ensures your audience membership is as current as necessary for your business use case.

mParticle offers two types of refresh frequencies:

1. Real-time – Audience membership is automatically updated as users either meet or no longer meet the audience criteria.
2. One-time – Audience membership is calculated only a single time, and does not refresh.

One-time Audiences

Setting an audience’s Refresh Frequency to Once creates a one-time audience: an audience that calculates its membership a single time and does not refresh on a recurring basis. One-time audiences are a good fit for large audiences built on long-term, historical data, where real-time or repeated updates aren’t necessary.

A one-time audience expires 30 days after the scheduled start date. Before it expires, you can continue to connect it to new outputs, and mParticle sends its already-calculated membership to each new connection (see Connect an Audience).

After a one-time audience expires:

• Its size shows as 0.
• It can no longer be downloaded.
• You can’t add new connections or enable Match Boost.
• You can’t edit its environment, refresh frequency, or starting date.

An expired audience displays the Expired status (see Audience Statuses). Expiration does not remove membership that was already written to user profiles, so a real-time audience that references the one-time audience’s membership is unaffected when the one-time audience expires.

Key Limitations

• Refresh type is permanent – You cannot switch an audience from real-time to once after creation. Delta processing – Only changes (users who newly qualify or no longer qualify) for realtime audiences are forwarded to connected outputs. For one-time audiences there will be no delta processing, because one-time audiences are only calculated a single time.
• Historical data loads - Data from historical data replays will not be reflected in one-time audiences. You must clone the audience and re-activate it to have the data from the historical load be available for audience processing.
• Audience groups can’t combine realtime and one-time audiences — You can’t create both realtime and one-time audiences within the same audience group.
• One-time audiences can’t reference realtime audiences - You can’t add membership criteria to a one-time audience that refers to membership of a realtime audience. The reverse is allowed: a realtime audience can reference membership of a onetime audience.
• One-time audiences must be the root audience of a group - A one-time audience can only be the top-level (root) audience in its group. It can’t be nested as a child audience within an audience group.

Audience Preview

After defining your audience criteria, you can use the Preview tab to inspect a sample of users that match your current segmentation. Audience Preview helps validate audience composition before activation, reducing errors and improving targeting accuracy.

How to Use Audience Preview

1. Navigate to the Preview tab in the audience editor.
2. View a list of sample users that meet your audience criteria:

3. Each row in the preview includes:

   • mParticle ID – A unique identifier for each sampled user.
   • Last Seen – The most recent interaction timestamp.
4. Click on a user’s mParticle ID to open their User Profile for deeper insights.

Audience Preview enables quick validation of audience logic, ensuring that the right users are included before activation.

Audience Insights

Audience Insights helps you better understand changes in an audience’s membership, the potential reach of an audience, attribute distribution across your audiences, and how audiences overlap.

For composable audiences, you can access Audience Insights directly from the Audience Builder.

For real-time audiences, you can access Audience Insights directly from Audience Builder. If a real-time audience has an Active or Calculated status, you can also open a dedicated Audience Insights page by selecting the Insights icon in its row on the Audiences landing page:

To learn more about Audience Insights, see Audience Insights.

Precise vs. Estimated Audience Sizes

Throughout the Audience creation process, you will see both preliminary and precise audience sizes.

Preliminary Estimates

Preliminary estimates are displayed before the audience has been fully calculated, and are denoted with a ~ character throughout the Audiences feature. When an Audience is created or edited, for example, preliminary estimates are shown to give an approximate idea of audience size.

Precise Estimates

Once an audience definition has been defined, a precise estimate will be displayed for that audience, even if it has not been explicitly activated or connected. It can take up to 15 minutes for the precise estimate to finish calculating. Waiting for the precise estimate is a more efficient option than activating the audience just to get a sense of its size. Precise estimates will display a cross-hairs icon wherever they are displayed throughout the segmentation experience:

Activate an Audience Without Connecting

The Activate Without Connecting feature allows you to manually activate an audience so that it calculates audience membership and updates user profiles, even if no outputs are connected. Common use cases for activating an audience without connecting it include:

• API-Access – Membership in an activated audience appears in user profiles and is retrievable via the Profile API.
• Testing and Troubleshooting – Before sending an audience to a downstream integration (e.g., Facebook or Braze), you may want to verify its size, composition, and updates to ensure accuracy and confidence in segmentation.
• Separation of Activation and Connection – You do not have to delete an audience to disconnect it from an output. This makes audiences more flexible and versatile, as you can preserve their definitions and criteria through the process of connecting and disconnecting outputs.
• Deactivate without Deleting – If an audience is no longer needed but it may be in the future, you can manually Deactivate it in the UI without deleting it. This preserves the audience definition for future use while stopping membership calculation and preventing further billing.

Audience Statuses

Audiences can exist in four states:

• Inactive – The default state when an audience is first created.
• Active – A manually activated audience, either through this feature or by connecting to an output. An audience with this state is actively calculating/processing audience membership changes.
• Calculated – A state where an audience is automatically activated by mParticle because other audiences depend on it. An audience with this state is actively calculating/processing audience membership changes. Calculated audiences are not charged from a billing perspective.
• Expired – Applies only to one-time audiences. A one-time audience automatically expires 30 days after the scheduled start date. After it expires, its size shows as 0, it can no longer be downloaded or connected to new outputs, and you can’t enable Match Boost or edit its environment, refresh frequency, or starting date. For more information, see one-time audiences.

Activate an audience: To activate an audience after it has been created (but before any outputs have been connected), click the three vertical dots in the top right of the audience tile, then select Activate.

Deactivate an audience: To deactivate an audience once it is activated, click the three vertical dots in the top right of the audience tile, then select Deactivate.

Considerations when activating audiences

• Billing Impact: Once an audience is activated and has an Active status, billing begins regardless of whether it is connected to an output or used for any specific action.
• A/B Audience Billing: Only audiences with an Active status consume a real-time credit, and an A/B test uses one credit total, no matter how many variations are active.
• Sibling cascade: If an audience in a branch moves to Active status, mParticle will automatically transition any dependent sibling and parent audiences into the Calculated state.

Activate an Audience in a Campaign

Once you have created an Audience that you want to forward to an external tool for use in a campaign, click the Connect Output button in the Audience tile, then follow the steps to connect that audience to any of your connected outputs.

Main Criteria Types

You can build criteria based on two main sources of data:

Events

Event criteria check for specific events and their properties, and their availability is subject to the data retention policy of your account. Within the new criteria option in the audience builder, the following options create event-based criteria:

• Events
• Ecommerce
• Crashes
• Installs
• Uninstalls
• Sessions
• Upgrades
• Screen views

User Profiles

These criteria check your active user profiles, and their availability is subject to the user profile retention policies of your account. Within the new criteria option in the audience builder, the following options create profile-based criteria:

• Users: Access user profile information such as user attributes, calculated attributes, current audience memberships, consent state, location, etc.
• User Predictions: Predictive Attributes that you have defined and generated for your users.
• Attribution: Access user install and uninstall information to build criteria based on the attributed campaign and publisher.

Circular references in audience membership conditions

You can add criteria to an audience definition based on audience membership. For example, if you have two audiences, A and B, you can specify that if a user is a member of audience A, they should be included in audience B.

However, the Audience Builder will not allow you to add audience membership-based criteria that results in a circular reference. For example, imagine the following scenario:

• You’re editing an audience called High-Value Shoppers.
• You want to add the audience membership condition: “User is a member of Holiday Campaign Responders”.
• But Holiday Campaign Responders already references Mobile App Users, and Mobile App Users includes High-Value Shoppers.

In this case, High-Value Shoppers ends up referencing itself indirectly through the other two audiences, making the definition invalid.

To prevent this, the Audience Builder would exclude Holiday Campaign Responders from the list of audiences in the dropdown menu. If you notice that some of your audiences are not displayed when creating membership-based conditions, this could be why.

Data Types and Matching Rules

Audience criteria can be created with several different data types, each with its own matching rules.

Strings

When building audiences based on string attributes, several case-insensitive matching rules can be applied:

• Contains / Does Not Contain: Matches substrings. For example, "blue" matches both "blue" and "blue shirt".
• Exact Match / Does Not Match: Requires the entire string to match. For example, "blue" matches "blue", but not "blue shirt".
• Pattern: Allows wildcard matching. * represents any number of characters, and ? represents any single character. For example, "bl?e" or "b*e" would both match "blue".
• Includes / Does Not Include: Matches an exact value within a list. For example, specifying "Chicago" in a list of movies returns "Chicago (2002)" and "Chicago (1927)", but not "Chicago Cubs".
• Partial Match: The inverse of Includes. For example, specifying "Chicago" would return all movies with "Chicago" in the title.

Date-Based Criteria

Filters based on fixed calendar dates. For example, events occurring after 09/12/2018. Date-based criteria are defined in UTC and are not relative to when the audience is calculated:

• Before: Excludes the specified date. For example, “Before September 9, 2018” includes events up to 11:59 PM UTC on September 8.
• After: Includes the specified date. For example, “After September 9, 2018” includes events from 12:00 AM UTC on September 9.
• On Date: Covers events from 12:00 AM to 11:59 PM UTC on the specified date.
• Between Dates: Includes events from the start of the first date to the end of the second date.

Recency-Based Criteria

Defines a period relative to the current time. For example, users active within the last 7 days. Recency-based criteria select events occurring within a timeframe relative to ‘now’.

• For example, if calculated at 1:00 PM on September 9, 2018, a recency filter for the last 7 days includes events between 1:00 PM on September 3 and 1:00 PM on September 9.

Attribution

Attribution criteria segment users based on campaign interactions, such as app installs or re-engagements.

• Profile Criteria: Select Attribution, then choose Install or Uninstall to filter based on campaign and publisher fields from attribution events.
• Event Criteria: Select Events > Attribution to use install or engagement events, filtering by event attributes. This allows you to select any information included with the event as custom_attributes.

Note: All event criteria are subject to audience event retention limits.

Identity

Identity criteria segment users based on their stored identities. You can test for the existence of a specific identity or apply string-based logic. These criteria are scoped to the workspace in which the audience is created. For example, if your account has three workspaces, an audience in one workspace only includes users active in that workspace.

Location

Location criteria allow segmentation based on geographic information.

• Equals: Filter users in a specific city, state, zip code, or DMA (Designated Market Area), based on IP geolocation.
• Within: Filter users within a set distance of a global city using latitude and longitude coordinates.

Cart

For ecommerce events, you can target users who added items to their cart but did not complete a purchase.

• Cart Abandonment: Select New Criteria > Ecommerce > Shopping - Cart Level > Cart Abandonment to define this segment.
• Specify the time period to wait without a purchase before including users in this audience.

Attribute Key

Use Exists or Not Exists to check for the presence of an attribute.

• For example, User Attribute: Gender EXISTS evaluates as true for both Gender = "Female" and Gender = undefined.

Additional Audience Features

Hybrid Audiences

Hybrid Audiences allows you to combine a composable audience that runs in your warehouse with real-time criteria in the Audience Builder. At a high level, you start with a composable audience that has been enabled for Hybrid Audiences, then reference that audience as part of your real-time criteria.

When a composable audience is enabled for Hybrid Audiences, mParticle matches each warehouse user to their mParticle profile using the identities configured in the underlying data model (such as email or customer ID). For each user where a match is found, mParticle records that user as a member of the audience on their profile. This is what makes composable audience membership available as a criterion in the Audience Builder.

Before you begin, make sure that:

• Your composable audience is built on a user data model where Enable this Data Model for Hybrid Audiences is turned on. This tells mParticle which identities to use when looking up mParticle profiles for warehouse users.
• The composable audience has Audience Membership Settings set to Enable for Hybrid Audiences, and the audience has been activated so that membership is written to user profiles.

To create a hybrid audience in the Audience Builder:

1. Follow the steps above to create a real-time audience and open the Audience Builder.
2. Click Add criteria, then choose Users to add profile-based criteria.
3. In the criteria editor, select Audience Membership and choose the composable audience you want to reference. This limits your audience to users who are also members (or not) of that composable audience.
4. Add additional criteria as needed, such as recent events, recency windows, or user attributes, to refine your real-time logic.
5. Click Done when you have finished adding criteria, then continue configuring and activating the audience as you would for any other real-time audience.

Hybrid audiences you create this way are activated and connected to outputs in the same way as other real-time audiences. You can learn more about hybrid audiences in the Composable Audiences Overview.

Auto-complete

As you define your audience criteria, a list of suggested matching values will appear based on what you’ve entered.

This feature works both when building new audiences and fine-tuning existing ones, helping you save time, reduce manual effort, and improves accuracy. To use it, you must have one of the following standard Roles: User, Admin, Audiences-only, Support, or Admin+Compliance. Alternatively, you can create a Custom Role with any of the following tasks: audiences:draft, audiences:edit, catalog, or audiences.

Boolean operators

Once you have added criteria, you can use the Boolean operators And, Or, and Exclude to create logical relationships with subsequent criteria.

Audience Environment Considerations

Be mindful of your selected audience environment:

• Use Production for real customer data to prevent test data from overwriting profiles in partner systems.
• Ensure parent and child audiences share the same environment. If environments differ, child audiences will not inherit users from their parent.

Audiences in Development display a badge in the Audience Group Editor, while Production audiences do not.

By following these steps and best practices, you can build and activate audiences that align with your business objectives, enabling more targeted user engagement and monetization strategies.