Personalized site search

Standard search matches words. It does not know that two shoppers typing the same word want different things. Personalized search adds what the store knows about the person to the ranking: the sizes she buys, the brands she clicks, the price range she stays in. It also handles what plain search gets wrong: typos, synonyms, a second language.

Why it matters for a store

Shoppers who search are a minority, about 1 in 6 by the public figures we cite, but they are the ones who know what they want. In a kids' fashion store the 4.9% of visitors who searched produced 27.5% of revenue, with a buyer rate 4.5 times the rest, as the case study on our site shows. Getting the first screen right for those shoppers is worth more than any banner.

An example

A shopper types "run" on a sports store. Before she finishes the word, the box suggests trail runners in the size she bought last time and the brand she clicked yesterday, ahead of the running socks the plain index would have listed first.

How the agent uses it

Search is one of the jobs of the sales agent, on the same memory as recommendations and chat. The team keeps control with merchandising rules on top of the ranking.

Related terms

See also Zero-results search, Merchandising rules (boost and bury), Anonymous visitor.

On the site

Search-as-you-type on the web and in the app. Results follow the live catalog, the sizes and categories the shopper returns to, and what similar shoppers bought. Every search joins the shopper's profile, so product pages, chat and the next email know what they looked for.