2 min read

Query Suggestions overview

The best friend who always ends your sentences

List of predictions that display whenever the shoppers type, designed to help them to complete and nail down the query. This feature not only minimizes having to write the whole query, but it also helps you to show shoppers what to look for to get relevant results. What’s more, it’s a great way of providing deeper access to the product catalogue.

Query Suggestions

Spot the difference

Do not mix up Query Suggestions with query results. Query Suggestions are hints to improve query formulation, guaranteeing that shoppers get accurate results.

For example, shirt dress, knit dress, or maxi dress can be suggestions for the query “dress”, bringing out more specific and narrowed results.


Remember that Query Suggestions appear when shoppers interact with the search box. But there are other types of suggestions that can display hints based on trends, search history, or product references before shoppers have started to type.

Try Query Suggestions to...

  • Anticipate shopper intent using machine learning technology.
  • Speed up query entry, especially for mobile devices.
  • Sharpen query formulation. Educate shoppers on how to perform better queries.
  • Improve product findability. Guide shoppers to the products they want, even those they don’t know.
  • Add spell-check and case-sensitivity behavior.

The inner workings of Query Suggestions

Query Suggestions are based on shoppers’ actions in the search UI. As shoppers perform different searches in the commerce store, behavioral information (queries, clicks, add to cart, etc.) is collected using the Tagging microservices.

The Empathize batch process analyzes and processes the tagging information to generate query suggestions candidates. For example, sweater, sweatshirt, sweat vest, and sweat dress are identified as query suggestions for the query “sweat”. Then a review process is performed to check whether candidates are valid according to different filters (e.g. they don't feature on a blacklist, they aren't duplicated, etc.), and eventually organize them by popularity.

After a number of days of data collection, the Empathize batch process calls the Index microservice to index the processed query suggestion candidates in the search engine and make them available as query suggestion candidates.

At query time, Empathy Platform Interface X uses the Search microservice to retrieve the list of query suggestions that best meet the specific shopper's intention. In your commerce search, when the shopper starts typing a query, valid search suggestions are immediately displayed in the predictive layer so that the shopper can start navigating the catalogue even before the search is fully typed.


The suggestion candidates from the Empathize batch process are also used for generating Popular Searches.


For a correct performance, make sure that your current search service supports this type of feature.


Explore the interactive map to see how the Query Suggestions relates to the other Empathy Platform features and microservices.