Prediction templates

Predictions templates give you a fast starting point for the most common prediction use cases. Each template comes with preset configurations, so you can build and launch a model without starting from scratch. For a fully customizable option, see Custom predictions.

📘

To understand how to read your model's quality metrics and decision tree after building a model, see Interpret prediction results.

Purchase prediction

Purchase prediction identifies customers most likely to complete a conversion event in a chosen future window. Use it to choose the audience for more expensive campaigns — target only those with a high probability of conversion through paid channels.

Open email prediction

Open email prediction estimates how likely a customer is to open an email. Use it to decide which customers to include in your email sends — and to protect email deliverability by avoiding sends to customers unlikely to engage.

📘

Only map the status property of email events. Other properties, such as action_type, will cause the prediction to fail. This template requires email events with the status delivered to identify eligible customers — if no such events exist, the prediction won't function.

Optimal send time prediction

Optimal send time prediction predicts the hour when each customer is most likely to open or click an email, based on their past session_start behavior.

To put this prediction to use, see Optimal send time in scenarios and campaigns.

Churn prediction

Churn prediction identifies previously active customers likely to stop engaging within a chosen future window, so you can take action before they churn.

In-session prediction

In-session prediction identifies customers likely to purchase during their current session. Use it to trigger real-time actions—such as a web overlay or promotional offer—to convert them before the session ends."

Custom prediction

Custom prediction lets you define your own target and prediction type — probability, class, or value. Use it when none of the standard templates fit your use case. For example, predict the best channel for each customer to maximize campaign performance.