Myth #3: Default search engines are good enough.
Search engines are making it increasingly easy to create and deploy a search service with serviceable default configurations. These defaults may appear to work well for simple queries, but often fail on edge cases and complex queries. Tuning and customization to make search more domain specific is almost always required. One example of an advanced search query is:
Ketchup, no corn syrup
This search uses negation, meaning the shopper wants a product with the absence of something - in this case an ingredient. Commercial search engines struggle with negation detection, often returning the opposite results the shopper sought. Here’s a real life example from a major grocer:
Another example of a common search engine failure is numeric detection. When shoppers request an item with an associated number such as size, volume, number of items in a package, or weight, search engines often fail. For example, when we searched for a hiking pack under 3 lbs on a major sporting goods site, the search engine could not process the numeric value and returned a hiking sweater. Numeric detection is a complex, sophisticated search capability for search engines.
↑ This is a real search!
Rather than relying on the status quo, or out of the box search engines, retailers can and should up their game with a customized, AI-powered search engine. (And for the record, this does not have to be an expensive or time-consuming endeavor.)