Users expect suggestions as they type. This requires sub-ms responses.
Approach : Prefix matching. Store terms in a trie. "app" matches "apple", "application", "appetizer."
Approach : Popular queries. Cache top searches. "iph" suggests "iphone " because millions searched it.
Approach : Personalized. Weight by user history. If you buy books, "har" suggests "harry potter" first.
In interviews, discuss the trade-off between prefix tries (complete) and cached suggestions (fast).