Predictive audiences are dynamic user groups in mParticle that use machine learning to forecast future behaviors—such as the likelihood of making a purchase—based on historical event data. Instead of relying only on past actions, they assign each user a probability score and percentile ranking, allowing you to target audiences most likely to perform a desired action.
\nYou can create predictive audiences in Audiences.
\nBefore creating a prediction, your workspace must meet the Predictions setup requirements: product access enabled, the required Predictions event output and feed input configured, schema synchronization and setup verification completed, and at least 45 days of usable historical data. For more reliable results, 90 or more days of historical data is recommended when available.
\nIf setup is incomplete or you don’t have enough historical data, prediction creation isn’t available yet. Continue collecting data through the live integration or contact your mParticle account representative to discuss whether historical replay or setup support is available for your account. A replay must supply enough usable historical data before you can create a prediction.
\nFor connection setup instructions, see Cortex event output and Cortex feed input.
\nTo create a predictive audience:
\nIf you haven’t already created a user prediction, click User Prediction from the list of criteria types and enter the User Prediction criteria before clicking Create Prediction.
\n![\"user](\"/images/audiences/using-predictive-audiences-1.png\")
If you’ve already created a user prediction:
\nAfter creating a predictive audience, a pipeline is created within Cortex, and the pipeline begins calculating. Calculation may take up to 24 hours, though typically the delay is about an hour.
\nIf there’s not enough data to learn from, the calculation may fail and the prediction won’t exist, resulting in an empty audience. To correct this, choose different criteria or troubleshoot your data for issues.
\nTo check on the status of your pipeline, view it in Cortex.
\nWhen setting criteria for a new user prediction, you can specify whether Cortex should generate that prediction for all users or a specific subset. Narrowing predictions to a subset of users can help improve the accuracy of your predictions, and avoid generating predictive attributes for users who are not relevant to a specific use case.
\nTo focus on a subset of users, select the option A subset of users in the Make predictions for field.
\n![\"Select](\"/images/guides/personalization/predictive-audiences/using-predictive-audiences/select-user-subset.png\")
Once you’ve selected this option, you can build queries with user attributes and behavioral events to select the users for whom Cortex will generate this prediction.
\n![\"Set](\"/images/guides/personalization/predictive-audiences/using-predictive-audiences/set-user-criteria.png\")
While it may be counterintuitive, using more data to generate a prediction does not mean the prediction will be more accurate. Not all users are relevant to every prediction you want to create, and irrelevant users can make it more difficult for the ML model to identify meaningful patterns and trends. This is why it’s best to consider the business outcome you want to achieve when defining the user segment that will generate a prediction.
\nObjective: Non-subscriber to subscriber conversion
\nObjective: Subscription upgrade (tier 1 to tier 2)
\nObjective: Churn prevention
\nObjective: Cross sell: Get purchasers of item A to buy a variant at a higher price point
\nObjective: Predicting which viewers who have not yet watched a show will watch it in the future
\n