Session-based recommendation selects items from interactions occurring during a user's current session. It is most useful when a user has some activity but limited long-term history, providing immediate behavioral signal before collaborative filtering becomes reliable. section 4.7
Use session-based recommendation to:
- Blend current-session behavior with popularity and content-based filtering for users who have only a handful of interactions. section 4.7
- Encourage discovery across categories instead of narrowing recommendations too early. section 4.7
- Treat collaborative filtering cautiously until the user's history provides enough signal. section 4.7
Figure 4.19 The different stages of a user's lifecycle, and which types of recommendations can be used at each stage.

Use the lifecycle figure to place session-based recommendation alongside other methods across stages of user activity. section 4.7
