Why Fabric Data Models Drift — and Why DAX Can’t Save Them

In this episode, we explore why many data teams mistakenly treat their data models as objective truth—and how this misconception leads to flawed decision-making. The conversation dives into modern analytics stacks, the limitations of “fabric” or centralized data models, and why context, ownership, and intent matter just as much as the data itself. Key Themes & Topics

• The Myth of the “Single Source of Truth”
  • Why most teams over-trust their data models
  • How abstraction layers can hide assumptions and errors
  • The danger of treating derived metrics as facts
• Data Models Are Opinions
  • Every model reflects decisions made by humans
  • Business logic is embedded, not neutral
  • Analysts and engineers encode trade-offs—often implicitly
• Execution vs. Understanding
  • Data engines execute logic perfectly, even when the logic is wrong
  • Accuracy in computation does not equal correctness in meaning
  • Why dashboards can look “right” while still misleading teams
• Ownership and Accountability
  • Who actually owns metrics and definitions?
  • Problems caused by disconnected analytics and business teams
  • The need for shared responsibility across roles
• Context Is More Important Than Scale
  • More data does not automatically mean better decisions
  • Local knowledge often outperforms centralized abstraction
  • When simplifying data creates more confusion than clarity

Notable Insights

• Treating analytics outputs as facts removes healthy skepticism.
• Data platforms don’t create truth—they enforce consistency.
• Metrics without narrative and context are easy to misuse.
• Trust in data should be earned through transparency, not tooling.

Practical Takeaways

• Question how metrics are defined, not just how they’re calculated
• Document assumptions inside data models
• Encourage teams to challenge dashboards and reports
• Prioritize understanding over automation

Who This Episode Is For

• Data analysts and analytics engineers
• Product managers and business leaders
• Anyone working with dashboards, KPIs, or metrics
• Teams building or maintaining modern data stacks

Become a supporter of this podcast: https://www.spreaker.com/podcast/m365-fm-modern-work-security-and-productivity-with-microsoft-365--6704921/support.

If this clashes with how you’ve seen it play out, I’m always curious. I use LinkedIn for the back-and-forth.