Autocomplete (also called typeahead or instant suggestions) reduces the friction of searching by predicting the most likely query completions as a shopper types. Good autocomplete pulls from at least three sources: previously successful searches across the store, current product titles and categories, and a popularity signal so trending queries appear first.
On ecommerce sites, autocomplete materially affects conversion: shoppers who interact with a suggestion convert 2–6× more often than shoppers who type the full query themselves. The mechanic also lets stores recover from typos before they happen — a half-typed “sneek” can still surface “sneakers” if your suggestion index is fuzzy-tolerant.
Common gotchas: serving stale suggestions after a catalog refresh, ignoring per-language indexes for multilingual stores, and over-indexing on raw popularity (which buries newer products). Most search platforms now combine BM25 lexical matching with embedding similarity so semantic near-misses (“running shoes” → “trainers”) still surface.