What ongoing split is recommended for site traffic to help Automated Personalization learn and adapt over time?

The recommended ongoing split for site traffic that helps Automated Personalization learn and adapt over time is a random experience. Implementing a random experience allows Adobe Target's algorithms to introduce variations in content display to a portion of the audience. This randomness is crucial because it ensures that users are evenly exposed to different content variations, enabling the system to gather diverse interaction data.

As users interact with various site experiences, the automated personalization engine uses this real-time data to analyze preferences and behaviors, adjusting its recommendations accordingly. This method maximizes the opportunities for the system to learn about user preferences, delivering increasingly relevant content over time based on collective user behavior patterns.

In contrast, other options such as shared experience or mixed experience may not provide the necessary randomization for effective learning, as they could limit exposure to specific audiences or conditions that do not yield broad data insights.