Semantic Layer 101

How to standardize metric definitions across BI

One of the biggest problems faced by any data team is a lack of clearly defined metrics across business teams. I’ve found that metric definitions tend to be all over the place, each team with a different idea of what the same thing means.

One team defines a customer’s current plan as including trials.

Another team defines it as only paying customers.

This makes it impossible for a data team to deliver a data model with clear definitions and guidance on how to use the data.

Semantic layers were created to help solve this problem.

Believe it or not, semantic layers are nothing new. They’ve been around since dimensional modeling, yet seemed to be forgotten about until recently.

Semantic layers make it so business users can access the data they need without any of the complexities. No technical jargon or SQL code, just business-friendly vocabulary.

Semantic layers leave no room for debate over what calculation or technical definition to use because the data team has already done the work to define this across the entire company.

With all this being said, this post from

Joe Reis

a few months ago made me laugh 😬:

What is a semantic layer?

Semantic layers sit between data models that live in your data warehouse (the transformation layer) and BI dashboards and reports. They help business users make sense of the data in data models so that they can be used to communicate metrics in BI.