Personalization policy

Personalization aggressiveness: 50%

This case study models 2,000 shoppers browsing a catalog of 40 categories. More aggressive personalization narrows recommendations toward each shopper's known preferences, lifting short-term clicks but shrinking the range of categories shown — a narrower "filter bubble" that can suppress discovery and long-term catalog reach.

Session outcome

Click-through rate—

Distinct categories surfaced—

Long-term reach index—

Recommendations shown by category (bar height = share of impressions)