The Case for a Federated Data Team Structure

I’ve spent years leading data teams and helping companies figure out how to structure them.

At WeWork, I saw the tension between centralized and embedded models, always trying to balance speed with consistency. At IBM, I worked inside a centralized team, watching what happens when data is shared across business units but owned by no one.

I’ve seen both sides:

• Centralized teams that move too slowly and frustrate stakeholders.
• Decentralized teams that spin in circles with conflicting numbers and duplicated work.

There’s no perfect answer. Every company is different, and how to organize a data team depends on the stage, size, and needs of the business. But after seeing these patterns play out across different environments, I’ve come to believe that a hybrid approach, one that balances centralized governance with embedded expertise, tends to work best.

If you’re leading a data team or working with one, you’ve probably faced this frustration:

• The business team wants insights fast.
• The data team is buried in requests.
• No one’s really sure what’s getting prioritized.
• At the end of the day, the impact of data isn’t clear.

If this sounds familiar, you’re not alone.

Most Data Teams Are Either Too Slow or Too Scattered

Most data teams struggle because they lack a clear data team operating model, making scaling a data team effectively nearly impossible. When it goes wrong, it looks like this:

1. Sit in a silo, disconnected from the business, responding to endless requests, moving too slowly to make an impact, and delivering work that doesn’t always meet the need.
2. Are spread too thin, jumping from one function to another and never truly able to solve real problems.
3. Constant finger-pointing and duplicated work, different teams build different models to answer the same questions. No one agrees on the numbers. Everyone wastes time.

This tension mirrors a classic org design challenge: centralized vs. decentralized data teams. A fully centralized data team ensures consistency, governance, and efficiency, but risks being out of touch with business needs. A fully decentralized model places data experts directly within business functions, making them more responsive, but often leads to duplicated efforts, inconsistent reporting, and lack of alignment.

Neither extreme works well. A federated data team model provides a strong balance between standardization and business responsiveness.

The Federated Model: The Best of Both Worlds

I’m a big fan of the federated model. Here’s how it works:

1. The Central Team (the federation center)
   • Owns data infrastructure, pipelines, and governance (data warehouses, engineering, modeling, etc.).
   • Ensures data reliability, quality, and accessibility across the company.
2. Embedded Analysts & Data Scientists (the “federated members”)
   • Align directly with business functions (e.g., Sales, Marketing, Product, Finance).
   • Work closely with teams to understand real business needs and proactively deliver insights.

Why This Model Works

The federated structure works because it combines tight business alignment with centralized standards and tooling.

1. Tighter feedback loops: Business teams get insights tailored to their goals.
2. Better prioritization: Analysts focus on high-impact work, not random one-off asks.
3. Faster iteration: Instead of a slow queue, data professionals work with stakeholders in real time.

The result is less frustration, clearer priorities, and data that actually moves the business forward.

Think Like a Product Team, Not an IT Help Desk

Imagine a product team that never talks to customers.

They’d build the wrong features, spend time on the wrong problems, and ultimately fail to create value.

Data teams are no different.

• If you only react to requests, you’ll always be behind.
• If you don’t understand the business deeply, you won’t create real impact.

A great data team doesn’t just answer questions, they help shape the right questions in the first place. And the only way to get the appropriate level of context and alignment to shape the questions is to sit with the team.

Your Data Team Is a Scarce Resource. Spend It Wisely.

"But what if we’re understaffed?"

Good question.

Let’s say you have three data people supporting six business functions. What’s better?

👍🏻 Delivering high-impact work for three functions

👎🏻 Spreading thin and delivering mediocre results for all six

Half-baked analytics doesn’t help anyone. Aligning data teams with business goals is key, prioritize where data can drive the most business value and expand from there.

Data Teams Get Labeled as “Cost Centers” for One Reason: Structure

The way you structure your data team determines whether they’re seen as order-takers or strategic partners.

So, if your data strategy isn’t working, take a step back.

• Is your team embedded in business functions?
• Are they prioritizing impact, not just requests?
• Are they seen as a strategic driver, not a reporting factory?

If the answer isn’t clear, it’s time to rethink. A well-designed federated model can position your data team into a business driver, not just a line item.

‍