Articles in this section

Building a Predictive Audience

Building predictive audiences can be an iterative process. After you create an audience and review the resulting prediction, you might decide to edit your criteria to see if you get a different result. Use the steps in this article to create, edit, or delete a predictive audience.

# In this article

# Create a new predictive audience

NOTE

At this time, you can create up to six predictive audiences. If you already have six, and want to create another, delete one of your existing ones. See Delete a predictive audience.

To create a predictive audience:

  1. Navigate to Insights > Predictive Audiences.

    • Choose one of the Predictive Audience templates with predefined system events for a specific business audience. This can be a good option if you're just getting started with Predictive Audiences.

    • Or, click New Audience to build a predictive audience using criteria that you specify based on your particular business goals.

  2. On the Audience Criteria page, review the values provided for the template you selected or, if you're creating an audience from scratch, specify the events and properties that are relevant to your audience.

    For example, to find people who are likely to be highly engaged with your product, your criteria could look for users who might add at least one item to their carts OR redeem at least one promo code, AND who have a Premium account type within a given month.

    Define predictive audience criteria

    • Click Perform event or Have user field, depending on the type of data you’re using as criteria. For example, if you track the type of account a user has with an event called Premium, you'd select Perform event. However, if you track this information with a user profile field called Account that has a Premium value, you'd select Have user field.

    • Select events or user fields and frequency operators and values.

      To ensure that the criteria you're selecting will support the audience insights you hope to achieve, hover the value you're thinking about choosing to see how it's performed over the past 30 and 90 days. Remember that audience criteria that applies to most, or very few, of your users won't produce insightful predictions.

      Preview event of user field usage

    • If you want to include more audience criteria, add it:

      • Immediately following the first item to create an "or" statement (if either criteria is satisfied, a user might convert on the predictive audience).

      • In the box below the previous entry to create an "and" statement (this criteria must also be satisfied to count as a conversion).

  3. If you want insight into how many users have recently met the criteria you've defined before you create an audience, click Preview users.

    If significantly more users satisfied the criteria than you expected, you might want to refine your audience by adding additional criteria. If far fewer users satisfied the criteria than you expected, you may want to select fewer or different criteria. Making these types of changes now instead of waiting for a audience to build can save you valuable time.

  4. Under Details, provide a Predictive audience name and a User field name to display user-specific score that will appear in segmentation and on the user profile. Structure it as you do other metrics used in your project (for example, HighlyEngagedPropensity).

    Refer to View a user's conversion scores to see how the value you provide for User field name appears in a user profile.

  5. Click Create audience. After 12–24 hours, you'll be able to review results.

TIP

To learn how a sample brand created their first predictive audience, see Create a New Predictive Audience for C&M.

# Review your prediction

To view the results of a prediction that has a Ready status:

  1. Navigate to Insights > Predictive Audiences.

  2. Click the audience you want to view.

  3. Select a group of users with conversion probabilities you'd like to focus on. Use the radio buttons above the chart to select the top or bottom 10% of your users, or select a custom range by entering specific values (or using the sliders).

    View smaller user selection

    • If the prediction shows that most of the users in your project are likely to convert on a given predictive audience, edit your audience criteria to include additional criteria.

    • If the prediction shows that significantly fewer than 1,000 users are likely to convert, pare back your criteria.

    • If the prediction doesn't provide the insight you need, try different data points altogether.

TIP

To learn more about the information in your prediction, see Evaluating Your Prediction.

# Edit a predictive audience

To edit a predictive audience:

  1. Navigate to Insights > Predictive Audience.

  2. In list view on the Predictive Audience page, click the overflow button (three dots) in the audience's row, and click Edit.

  3. Provide the new details and click Update predictive audience. After 12–24 hours, you'll be able to review results.

# Delete a predictive audience

WARNING

Deleting a predictive audience also deletes the related conversion probability property and scores from all user profiles in your project. References to the conversion probability property (user profile field) that you've included in campaigns, journeys (workflows), lists, or templates will be impacted. Delete all references to conversion probability user profile fields before you delete an audience.

On the Predictive Audiences page, click the overflow button (three dots) in the audience's row, and click Delete. Within 12–24 hours, the predictive audience is deactivated, and all related user fields are removed from user profiles.

# Next steps

Related articles