Route search queries to query pipelines

This is for:

Developer

A query pipeline is a sequence of rules applied to search queries sent to Coveo. You can define many rules yourself (e.g., ranking expression rules, thesaurus rules, and filter rules). You can also associate Coveo Machine Learning (Coveo ML) models with a query pipeline and leverage Coveo’s AI.

The following diagram shows the order of execution of query pipeline features:

Order of execution of query pipeline features

You may want to create several query pipelines in your Coveo organization. Though there are different mechanisms to route your search queries to a query pipeline, we recommend using condition-based routing. Using this method, your query is evaluated against each query pipeline condition. Upon the first match, the associated query pipeline is applied.

Note

Using condition-based routing eases the comparison of query pipeline performance through A/B testing.

To leverage query pipeline features

  1. In your Coveo organization, create a query pipeline and condition.

  2. Validate that your search interface queries are routed to the new query pipeline.

    Tip

    Use the Relevance Inspector.

  3. Add query pipeline features (or components) to the query pipeline.

    Tip

    Though many query pipeline features let you alter search queries directly, leveraging Coveo machine learning is a better time and relevance investment. For example, Automatic Relevance Tuning (ART) and Query Suggestion (QS) models automatically improve search relevance based on actual user behavior and successful search outcomes.

  4. Repeat steps one to three to create other query pipelines and conditions as needed.