One Comment

  1. This makes sense for the cases you benchmarked, but the harder case is a user typing something like where their SKU never actually appears, say a plural or a slightly different word than what’s stored in the field. Does the relevance scoring have any fallback for near misses, or does it stay a strict match against whatever text made it into the index? That distinction seems like it’ll decide whether this feels like real search or just a faster version of the same LIKE query.

Leave a Reply

Your email address will not be published. Required fields are marked *