Centralized vs Decentralized Data Teams: What do Top Data Leaders Prefer?

In today’s data-driven world, organizations grapple with a pivotal decision: how to structure their data teams for maximum impact. Should decision-making power reside at the top in a centralized system, or should it flow freely across teams in a decentralized approach?

‍

Understanding the nuances, benefits, and challenges of centralized and decentralized models is essential for businesses striving to solve this problem and stay ahead.

‍

Insights shared by the data leaders at Freshworks, Samsara, and EQOM Group reveal that the answer isn’t black or white. Instead, it lies in the grey area of adaptation and balance.

What are centralized data teams?

A centralized data team functions as a single, cohesive entity where expertise is concentrated within a core group. Centralized teams are often associated with standardized processes, high-quality outputs, and efficient resource allocation. 

‍

This results in consistency, efficiency, and standardization. Sachin Mishra of Freshworks underscores the utility of centralization, particularly for organizations experiencing rapid growth. 

‍

“Data teams staying closer to the business for certain functions works well in the early days. As you scale up and start getting into operational excellence and other challenges – that’s the right time to think about some of the centralization and other pieces.”

~ Sachin Mishra, Senior Director of Data Science and AI, Freshworks
Leveraging AI to make data accessible across Freshworks

‍

This perspective is echoed by Irina of EQOM Group, who emphasizes how centralization fosters alignment across geographically diverse teams. 

‍

“The data team is centralized at EQOM Group. A lot of other services are centralized too. What helps us prioritize is our connectedness across global teams. The Dutch are more direct whereas the French are more diplomatic. Being centralized helps us align our priorities.”

~ Irina Ioana Brudaru, Principal Engineering and Science Lead, Data, EQOM Group
The swiss knife approach to data teams at EQOM Group

‍

When centralization works, it works well. Processes are smoother, data is cleaner, and collaboration flows effortlessly. But not all organizations can — or should — centralize everything.

‍

What are decentralized data teams?

In a decentralized model, teams work independently and enjoy more autonomy. They’re agile, adaptable, and capable of responding to unique challenges quickly.

‍

This model is particularly effective in dynamic environments where speed and adaptability are critical. Decentralized teams are well-positioned to drive innovation and tailor their efforts to unique business challenges without being constrained by centralized oversight.

‍

This is the reality at Samsara. Kiriti Manne, Head of Strategy & Data, explains their approach.

‍

"For company-wide data use cases, we centralize. For everything else, teams are autonomous. This balance ensures we innovate while staying aligned," 

~ Kiriti Manne, Head of Strategy & Data, Samsara
How Samsara’s Attribution Model Turns Data into Gold

‍

Real-world approaches: Insights from leading data leaders

From EQOM Group’s empowerment-driven decentralized model to Samsara’s delicate balance of control and collaboration, and Freshworks’ hybrid approach — each example offers valuable lessons for businesses seeking to optimize their data operations.

‍

EQOM Group: The power of decentralized ownership

Decentralized ownership is a game-changer in how organizations operate. It shifts authority from the few at the top to the many across the organization. 

‍

This change isn’t just structural, it’s philosophical. When decision-making power is shared, employees feel trusted. When they feel trusted, they act with purpose, take initiative, and innovate. 

‍

The power of decentralized ownership is best seen in the example of EQOM Group, where each team member is given the autonomy to become owners of the outcomes, not just cogs in a machine. 

‍

“There is a unique way of working in our data team that I have never seen in data teams before. We work like independent consultants at EQOM group and our boss doesn't tell us what to do. If somebody needs an analysis, they go ahead and do it. No questions asked.”

~ Irina Ioana Brudaru, Principal Engineering and Science Lead, Data, EQOM Group
The swiss knife approach to data teams at EQOM Group

‍

At EQOM Group, data teams move fast and tackle challenges head-on. A data analyst designs and deploys a solution without waiting for managerial approval as and when they notice a bottleneck in customer onboarding.

‍

This “Swiss knife” approach exemplifies how decentralized models when executed thoughtfully, can create versatile, self-reliant, and well-guarded data teams.

‍

Samsara: Balancing control with collaboration

Choosing between centralized and decentralized data teams is ultimately a balancing act. 

‍

Centralized teams excel but falter when agility is needed. On the other hand, decentralized teams thrive in fast-paced settings and risk inefficiencies and fragmentation.

‍

Kiriti from Samsara acknowledges these trade-offs and advocates the importance of fostering communication and collaboration, regardless of the model. 

‍

“Staying connected across all related functions that are not 100% related is challenging. We sometimes work in silos, duplicating efforts when existing solutions are available. It's important to pull our heads up every once in a while across all priorities to see what we can leverage as an organization. Staying connected is key to avoiding inefficiencies."

~ Kiriti Manne, Head of Strategy & Data, Samsara
How Samsara’s Attribution Model Turns Data into Gold
‍

‍

At Samsara, teams operate independently to maintain agility but come together to align on company goals. This balanced approach between autonomy and accountability fuels innovation.

‍

“We try to be as centralized or decentralized as needed. For data sources that have use cases across the company, not just my team, our central IT team is responsible for the standardization, pipelining, and governance, so that everyone has access to the same quality data.”

~ Kiriti Manne, Head of Strategy & Data, Samsara
How Samsara’s Attribution Model Turns Data into Gold

‍

Freshworks: Hybrid model for agility and accountability

Freshworks, known for its customer-centric innovation, takes the hybrid approach to new heights.

‍

For this people-first AI service provider, deciding between a centralized or decentralized data team model was not easy. By going hybrid, Freshworks chose the best of both worlds.

‍

"We still have a very distributed ecosystem that’s aligned for speed and agility. Ours is a very hub-and-spoke model where we have central teams as well as teams working at the edge to support their respective business functions."

‍~ Sachin Mishra, Senior Director of Data Science and AI, Freshworks
Leveraging AI to make data accessible across Freshworks
‍

On the trade-offs between centralization and decentralization, Sachin comments:

‍

“You can have higher quality by centralizing, but it becomes a little bit slower, and then with decentralized teams, there’s a lot of fragmentation, Ultimately, it’s about efficiency and how we can optimize the overall cost and scale with a lot of open-source flexibility." 

~ Sachin Mishra, Senior Director of Data Science and AI, Freshworks
Leveraging AI to make data accessible across Freshworks

‍

Having built data organizations from scratch to a few hundred people with AWS in the last 10 years, Sachin advises that the best way forward is to keep an open mind.

‍

“Unfortunately, there is no magic pill for choosing between centralized or decentralized models. As your organization grows, how you align yourself to your product lines, strategic goals, and market conditions will dictate whether centralized or decentralized models serve you better. What has worked for me is staying flexible and not over-committing to either of these models because as your business needs and overall landscape shifts, data teams also need to evolve.”

~ Sachin Mishra, Senior Director of Data Science and AI, Freshworks
Leveraging AI to make data accessible across Freshworks

‍

Mixed model approach and not overcommiting is a recipe for success

What works for a small, nimble company today might crumble under the weight of future requirements. Early on, decentralization can be a game-changer. But as a business scales, priorities shift and operational excellence becomes critical. That's when centralization starts to look appealing. 

‍

For those looking to bridge the gap, a hybrid approach provides a scalable, adaptable solution.

‍

Ultimately, the decision between centralized and decentralized data teams is not about choosing one over the other but about aligning the model with the organization's goals, scale, and complexity. 

‍

As the debate continues, one thing is certain: success lies not in the rigidity of a chosen model but in its ability to evolve with the organization. By understanding the strengths and limitations of each approach, businesses can design data strategies that adapt to their unique needs, ensuring they remain agile and effective as they grow.

‍