Predictive Attributes Overview

A Predictive Attribute is a special type of User Attribute that tells you how likely a user is to perform a specific action in the future. While most User Attributes describe static information about your users like age, location, or subscriber status, Predictive Attributes are dynamic and forward-looking, giving you a real-time snapshot of a user’s probability of taking an action that matters to your business.

Why use Predictive Attributes?

Traditionally, marketers and product managers look to user events like website visits, page views, and journey completions to build out customer segments. While manual segmentation can certainly be effective, it is subject to the limitations and inefficiencies of human decision making.

Predictive Attributes let you avoid the drawbacks of manual audience building. Once you define your desired conversion goal, Cortex’s Machine Learning models will analyze thousands of behavioral signals to determine which users are statistically most likely to convert, letting you accelerate decision making and execute campaigns with confidence.

Additionally, Predictive Attributes update automatically based on real-time customer behavior. Since your highest value customers are always changing, Cortex continuously recalculates Predictive Attributes to ensure that your campaigns stay focused on the users who most closely align with your campaign goals.

When to use Predictive Attributes

The most fruitful use cases for Predictive Attributes tend to have the following characteristics:

  • Behavioral events are an important factor in generating a prediction. User Attributes alone (e.g. Customer Lifetime Value, subscriber status, etc.) are not sufficient to predict user behavior.
  • The action occurs frequently. The more frequently an action occurs, the more behavioral data Cortex will likely have on the user base. (Like purchases and membership upgrades, for example.)
  • The action is intentional and meaningful. Conversion actions like purchasing an item, subscribing, and upselling reflect meaningful consumer decisions, whereas product views, page visits, and other browsing behavior can be less intentional.

Specific examples of use cases where Predictive Attributes are likely to improve campaign performance include:

  • Non-subscriber to subscriber conversion
  • Subscription tier increases
  • Upselling users on a premium offering
  • Churn prevention
  • Predicting media engagement (e.g. which users are likely to watch a particular show)

Was this page helpful?