The Old MarTech Stack Is Breaking, Part I: Semantic Layers Are the New Keeper of Data and Advertising with Leighton Welch & Sarah Martinez from Tracer

In this episode of Signal & Noise, we sit down with Leighton Welch (CTO) and Sarah Martinez (CCO) from Tracer to unpack a fundamental shift happening across enterprise data, marketing, and advertising:

The stack is being rebuilt around data—not applications.

And at the center of that rebuild? The semantic layer.

This conversation goes beyond the usual “data unification” talking points. Instead, we break down the five structural shifts reshaping enterprise data strategy right now—and why they matter more than ever as AI moves from experimentation to execution.

Tracer provides a practical lens into this transformation. Not as another dashboard or point solution, but as a system designed to solve a harder problem:

How do you create shared, trusted business logic across fragmented systems so both humans and AI can act on the same truth?

1. Why enterprise data strategy is suddenly back on the front burner
AI didn’t create the data problem—but it exposed it. As Leighton puts it, AI is forcing organizations to prioritize initiatives they should have tackled a decade ago, from centralized data ownership to consistent definitions. 

2. The five shifts redefining the stack

• Semantic layers as the new business logic layer
• Warehouse-native architecture replacing SaaS silos
• Zero-copy activation reducing data duplication
• Governance becoming infrastructure—not compliance
• AI readiness as the new forcing function

3. The semantic layer as the control plane for AI
If AI is the operating layer, then definitions, context, and trust become the system of control. Without a shared “data dictionary,” agents don’t just fail—they amplify inconsistency.

4. Why monolithic SaaS is under pressure
Enterprises are moving away from copying data into every tool. The warehouse is becoming the system of record, and everything else is being forced to justify its existence.

5. The real problem: not data, but meaning
Most companies don’t lack data—they lack agreement on what that data actually means. Tracer’s approach focuses on turning raw data into reusable, governed business logic that can power decisions across teams.

6. AI readiness isn’t about models—it’s about foundations
Clean data, consistent taxonomy, shared definitions, and governance aren’t “nice to have.” They are prerequisites. Without them, AI becomes a very fast way to make very bad decisions.

For years, the industry debated identity vs. platforms, CDPs vs. composability, centralization vs. federation.

But that’s not the real shift.

The real shift is this: Data is no longer an input to the system.
It is the system.

And the companies that win won’t be the ones with the most data—they’ll be the ones with the most trusted, reusable, and operationalized context.

• CMOs and marketing leaders trying to make AI real
• Data and analytics teams dealing with fragmented stacks
• AdTech/MarTech operators navigating warehouse-native architectures
• Anyone tired of waiting 6 months for insights that are outdated on arrival

If you take one thing from this episode, it’s this:

AI doesn’t fix bad data. It exposes it.
And the semantic layer is how you fix it.