A few weeks ago, I gave a talk at the Tableau Conference in San Diego, followed by a live webinar. This post captures everything I shared about a challenge data teams face every day: navigating the delicate balance between freedom and governance. Tableau or dbt? The answer isn’t one or the other. It’s both.
Freedom or Governance?
To get to the heart of the issue, let me start with a parenting story. I live in Tel Aviv, and when I traveled to San Diego for the conference, I had to leave my three kids. My two younger boys went to stay with their grandparents, but my 16-year-old daughter insisted on staying home alone for five days.
As a parent, this sparked a dilemma you might recognize: the tension between giving freedom and enforcing governance. On the one hand, we want our children to be independent—to explore, to make mistakes, to learn. On the other hand… five days alone? Who knows what might happen?
This exact tension—freedom vs. governance—is something we constantly wrestle with in our data work too.
Tableau vs. dbt: The False Choice
Let’s talk Tableau and dbt. Tableau represents freedom. It’s flexible, fast, and beloved by business analysts. dbt represents governance. It’s the industry standard for managing business logic within the data warehouse, version-controlled and optimized.
This polarity often presents a false choice: do we lock things down and ensure trust and consistency, or do we enable everyone to participate and move fast?
We need both. That’s why I co-founded Euno two years ago: to help data teams achieve exactly that balance. And I’m joined in this webinar by Josh Vitello, who spent over 16 years at Tableau helping customers scale self-serve analytics while wrestling with the consequences of data democratization.
The Reality: Everyone’s Struggling
In the last two years, I’ve spoken with over 400 data leaders. They all share a common vision: one central place to manage business logic and definitions. Many are using dbt to get there—and they’re excited about it. But almost every team also describes similar frustrations.
Some feel that dbt has slowed them down. Modeling in the data layer means everything has to go through a central engineering team. Business teams lose autonomy and agility.
Others give analysts too much access to dbt, only to find themselves in chaos—conflicting definitions, duplicate logic, and a mess of ungoverned views.
And some teams suffer from both issues at once.
Why? Because business logic evolves constantly. It’s not a one-time project—it’s a continuous process, driven by the business side, not back-office engineers.
The Classic Problems Persist
Even when you build your entire model in dbt, the classic problems remain. I like to categorize them into the three S’s: Scale, Speed, and Silos.
Scale: dbt requires an engineering mindset. Equipping analysts with dbt access starts with excitement, but eventually they’re pulled back into business needs and can’t sustain the engineering work.
Speed: Business needs fast insights. Engineering workflows slow things down. Analysts create workarounds—in Tableau, in SQL, in Excel—to keep up.
Silos: As the operation scales, central teams can’t keep up with modeling everything. Analysts revert to doing things their own way. And the silos, inconsistencies, and duplicates creep right back in.
Many large companies end up taking dbt away from analysts and putting it back in the hands of engineers, undoing the democratization they initially aimed for.
Meanwhile, Tableau becomes cluttered with legacy logic—custom SQL, calculated fields, and no clear ownership or history. This “dark side of Tableau” is risky to clean and impossible to fully understand.
The Solution: Visibility, Not Control
We don’t need to choose between Tableau and dbt. We need visibility into how logic created in Tableau maps to dbt and the ability to decide what’s worth formalizing.
The reality is that 90% of business logic created by analysts is never reused. It’s exploratory, one-off work. Over 80% of Tableau refresh jobs we’ve measured have zero utilization downstream.
To separate the signal from the noise, we need to connect utilization with lineage. The logic that’s valuable is the logic that’s used. The rest? Clean it up.
Metadata: The Key to Everything
All of this comes down to metadata, but not just any metadata. To make governance work, metadata must be:
Deep (not just surface-level)
Connected (across systems and layers)
Contextual (understanding dependencies)
Structured (clean and enriched)
Flexible (adaptable to how your org works)
That’s what Euno does. We transform metadata into an actionable foundation for governance and AI.
Our Approach: Map, Enrich, Activate
Map: We stitch together lineage and utilization at the column level—from the data layer to Tableau. All metadata is stored in a graph database, searchable via natural language.
Enrich: Define live metadata properties (e.g., “certified dashboards”) based on your organization’s rules. These properties update automatically when lineage changes.
Activate: Use metadata to trigger real workflows—optimize costs, prioritize modeling, align metrics, or certify AI-ready assets.
Immediate and Measurable Impact
Within hours of installing Euno, you’ll see insights like:
The cost of unused Tableau extracts still running warehouse queries.
Which high-use dashboards aren’t connected to dbt.
Which metrics in Tableau are duplicated or conflicted.
From there, you can trigger automated workflows—like archiving unused data sources, generating Jira tickets to prioritize certified dashboards, or resolving metric conflicts before adding them to a governed semantic layer.
Why This Matters for AI
Everyone wants AI-powered analytics: text-to-SQL, text-to-report, natural language interfaces. But without a governed semantic model, it won’t work.
LLM tools can’t guess which logic is correct. If you expose them to a chaotic environment with duplicates and inconsistencies, they’ll fail.
Euno helps by:
Decluttering your environment
Aligning and certifying key metrics
Defining and labeling what’s “AI-ready”
That way, you can confidently deploy AI agents only on trusted, documented, governed data assets.
How to Get Started
Implementing Euno is fast. One hour to install, a few more to map your metadata, and you’ll start seeing value within days.
We recommend starting with one or two core use cases—decluttering, cost optimization, metric alignment—and demonstrating real value in your evaluation. This isn’t a multi-year cataloging project. It’s a tangible solution for real data problems.
Final Thoughts
Whether you’re a parent, a data leader, or both—the lesson is the same. Freedom and governance are not opposing forces. You need both. You need balance.
Governance doesn’t have to be scary or restrictive. When powered by actionable metadata, it can be empowering, enabling your analysts to move fast while maintaining trust, consistency, and readiness for AI.
We’re here to help you get there.
