What Is Cortex Search in Snowflake?

At a high level, Cortex Search works by creating a searchable representation of your data and using that representation to find relevant information when a query is submitted.

The basic flow can be understood like this:

Source Data → Cortex Search Service → User Query → Relevant Results

For example, suppose an organization stores customer support information in Snowflake.

A customer asks:

"Why was my payment declined?"

Instead of searching only for the exact phrase "payment declined," the search system can identify related content about failed transactions, rejected payments, payment issues, or similar support information.

This makes the search experience more useful, especially when users don't know the exact wording used in the underlying data.

Keyword Search vs Semantic Search

One of the important concepts behind Cortex Search is the difference between keyword-based and meaning-based search.

Keyword Search

Keyword search looks for matching words.