I was drowning in meeting invites that Tuesday morning. Back to back calls about our Tableau migration, each one more urgent than the last. The irony? A week earlier, company executives sent a company-wide email asking us to “reduce time spent in meetings to 4 hours weekly”. I had more than 4 hours of meetings just that day!
But this was not just another tech migration gone sideways. We were chasing down a regulatory response deadline, and the IT solutions had just delivered numbers which were off by $120 million from what we had manually checked.
When Technical Excellence Misses the Point
Here’s what happened: Our team was migrating from Power BI to Tableau and had taken complete ownership of the BI solutions. They were great, technically flawless execution. However, they had one small problem. They defined dormant accounts using the old Power BI policy, completely ignoring new regulations that changed the dormancy period definition.
For context, dormant accounts in banking are not just forgotten savings accounts. After a certain period of inactivity, these accounts go through a process called escheatment (I know, I am glad I have to write it and not say it out loud). What escheatment does is basically closes the account and the remaining funds go to the state treasury. Regulators watch this process like hawks.
If we had submitted those numbers to regulators and discovered the error later? Most likely another 8 digits fine minimum. And the way it works in banking, if you were told to fix it once and you go back with issues still, the fines just keep climbing.
The Real Choice Every Organization Faces
This disaster (which we thankfully caught in those marathon meetings) highlights the fundamental decision every organization has to make:
Option 1: Enterprise Center of Excellence (CoE)
IT controlls everything
Single source of truth
Better governance
Four months of wait time for simple report modification
Option 2: Federated CoE
Business users control their analytics
Fast, responsive, tailored to business needs
Everyone delivers their own version of “truth”
Most executives treat this as a technical decision. I have learned that it is actually a business decision. It is about how fast you want to move versus how much chaos can you tolerate?
If you find this valuable, consider subscribing
Why IT Departments Hate Federated CoE (BI)
There are plenty of reasons, but the most obvious one? It makes their job harder. But keep in mind, “harder for IT” does not necessarily mean “bad for business”.
In a centralized model:
IT controls the entire tech stack
All requests flow through established processes
Career advancement comes from technical sophistication
In a federated model:
IT becomes a service provider, not a gatekeeper
Business users make mistakes (requiring IT to fix them)
Success gets measured by business outcomes, not technical excellence
The truth? Federated models deliver better business outcomes, partially because they are messier and less controller. The messiness gives teams the ability to tweak solutions based on their specific needs and KPIs.
Framework: Controlled Chaos
Before IT folks start sharpening their pitchforks, let me be clear: You do not have to choose between total control and complete chaos. The answer is Controlled Chaos, a three layer framework:
Layer 1 - Core Truth (Centrally controlled) - Things like customer definitions, financial metrics, regulatory reporting, and master data management. IT owns this completely. These are enterprise-wide metrics that must use identical defintions across the organization.
Layer 2 - Business Context (Federated with guidelines) - Department specific KPIs, operational metrics, project analysis, risk management approaches. These are useful only for specific business units, but they need guardrails.
Layer 3 - Exploration Zone (Completely open) - Ad-hoc analysis, experimental metrics, sandbox enviornments. Let people play and discover.
Implementation Rules That Actually Work
Golden Source Metrics: Define company wide metrics that must use the same definition everywhere
Self-service Boundaries: Clear rules on what business users can modify vs. what requires IT involvement
Quality Thresholds: Alert systems when federated reports deviate significantly from enteprise metrics.
Usage Governance: Monthly reviews of what gets built and what gets used (so we don’t end up with 47 dashboards that nobody opens, if you have not recent article I shared on this, you can check it - The $200k Dashboard That Nobody Trusted)
What This Means for Your Career
Whether your organization chooses centralized or Federated CoE (BI), your value comes from understanding both approaches and knowing when to apply each one.
The professionals who survive and thrive are the ones who can:
Translate between business needs and technical constraints
Design governance frameworks that enable speed without creating chaos
Measure success by business impact, not technical excellence
Adapt their approach based on organizational reality
The Three Page Cheat Sheet
A friend of mine recently took a role as Head of Data Tooling and BI at a local bank. They are implementing a federated CoE approach, and he was just sharing this information with me. I put together a quick two-page cheatsheet covering the key considerations for avoiding pain later.
It is a bit messy (I threw it together quickly), but it covers the essential questions you should ask before diving into federated BI implementation.
