Semantic layer

As your data team grows, a semantic layer becomes a useful abstraction to help you to create a consistent and unified view of your key metrics across the entire organization. It ensures that every analysis your do uses the same definitions when calculating complex metrics (e.g. How many users do we have? What was our revenue in the last fiscal year?).

Deepnote is compatible with all major approaches to building a semantic layer. In this article, we will cover different approaches and how to integrate them with Deepnote.

Building a semantic layer in a notebook

Deepnote is a great place to define your metrics. This is the recommended approach for smaller teams, or when your entire data stack is based on Deepnote.

We recommend creating a separete notebook for every metric definition. Using SQL or Python blocks to retrieve the data and text blocks to document each metric and history of changes in natural language.

In subsequent analyses, you can reference these canonical notebooks, or use API to retrieve the data from outside Deepnote.

Since these notebooks are in your workspace, Deepnote AI will pick up these definitions automatically and use them to answer your questions.

dbt Semantic Layer

You can use dbt or any other data warehouse-based semantic layers directly from Deepnote.

Read more about our integration with dbt here.

Looker and LookML

Looker exposes a JDBC interface, allowing you to query LookML models via SQL.

Get in touch with us at sales@deepnote.com to learn more.