The value of a metrics/semantic layer is so much more about organizational productivity and enablement than pure governance for data or analytics engineering teams.

In fact, I would go as far as to argue that data and analytics engineers should view a metrics/semantic layer as a mechanism to empower business self-service capabilities, rather than an internal tool for themselves to define and manage metric pipelines.

But, how do we live up to this promise?

Businesses are intricate input-output systems, and metrics/semantic layers treat business metrics as fundamental concepts in code. The current use cases involve asking this layer to consistently and reliably generate metric calculations - integrating in some capacity with BI/visualization tools.

However, while this reliability is valuable, considering the metrics layer as a foundational building block for other software to operate on opens up entirely new possibilities for working with data, some of which were previously unimaginable.

One exciting application, which we at Trace are passionate about, is Metric Trees. With these metrics building blocks, you can construct a metric tree that models entire business processes.

Equipped with rich metadata, a new application can automatically traverse this tree, seamlessly perform calculations, and generate valuable insights almost instantly.

Tasks that were once painstakingly hand-crafted, taking analysts days or even weeks, can now be executed within a matter of minutes. Requests for data or analysis from a product manager or a growth lead that would take weeks to be fulfilled are now readily available at their fingertips with just a few clicks.

This directly drives org-wide productivity and data ROI because today in any organization,  the number of analysts who understand both the data and SQL nuances, and also have a clear mental model of how the business operates, is a small elite group.

I envision a future where business strategy and entire operations can be streamlined by software built upon the foundations laid by metrics/semantic layers.

So, in my view, the “one-source-of-truth” value proposition for metrics layers fails to adequately capture how they can empower organization-wide self-serve data and analytic capabilities.

This is why I was so excited to see Tristan’s recap from DBT’s Coalesce conference that the  Semantic Layer workshop was sold out twice over!

So, whether you implement this using DBT or even start by capturing and organizing your metrics in code, the possibilities are exciting.

Let’s go!