Personalization

Tailoring search results to the individual shopper based on their history, behavior, or declared preferences — a per-user re-ranking layer on top of relevance.

Personalization signals come in tiers. Session-level: what the shopper has clicked, searched, and viewed in this visit. User-level: their past purchases, returns, and saved items. Cohort-level: the behavior of similar shoppers (collaborative filtering). Personalization is usually a re-rank stage that adjusts the relevance ordering using a learned user-item affinity score.

For ecommerce, the wins compound at the long-tail of users. New shoppers get cohort-based defaults; repeat shoppers get history-aware results. Personalization done right lifts conversion 5–15% but adds complexity to A/B testing (per-user variance becomes the dominant noise term).

Privacy and explainability matter: opt-out controls, no personalization on logged-out browsing where signals are noisy, and avoiding filter bubbles where personalization narrows discovery. Always offer a way to see “popular” unfiltered results.

Related terms