A non-personalized recommendation is the same product suggestion for every customer, without using an individual's purchases or profile. It suits new websites and new customers when personal data is unavailable. section 5.1
A practical recommendation workflow uses shared purchase behavior:
- Collect products frequently bought together rather than modeling individual preferences. section 5.1
- Suggest another product from the relevant group, using hardcoded associations when no purchase data exists. section 5.1
Cart-based suggestions
A customer adding a football can receive skis as a follow-up suggestion; other observed associations include milk with bread, laptops with external disks, and tennis rackets with tennis balls. section 5.1
