Centralised vs Decentralised Data Teams: Finding the Right Fit

I talk to businesses regularly about their data strategy and data team structure.

Team structure is hidden battle organisations fight as they grow their data teams.

Typically, data teams start centralised.

Then they decentralise. Then centralise. Then decentralise. Then settle somewhere in the middle.

And before you know it, there are Power BI Bandits across the business!

Slinging data and creating value...but not potentially not adhering to best practice.

So what's the go with decentralised vs. centralised data teams? What even is it?

Let's find out.

Understanding the Models

In the realm of data team structures, organisations often grapple with choosing between centralised and decentralised models.

A centralised model consolidates data professionals into a single team that serves the entire organisation. Stakeholders may then come to the central team and request projects or insights.

Centralised Model @dbt

A decentralised model embeds data professionals within individual departments, allowing for domain-specific focus, giving rise to titles like; Marketing Analyst, Finance Analyst, HR Data Analyst. Typically, the Data Engineering or Data Platform team may maintain a central governance and authority over the data sources.

Decentralised or Embeded Model @dbt

Many organisations adopt a hybrid approach, blending elements of both to suit their needs.

It's more complicated than this but at a high-level this is decentralised vs. centralised.

Pros and Cons

Centralised Model:

Pros:

• Consistency and alignment in data practices and governance.

• Facilitates resource sharing and collaboration among data professionals (mentorship).

• Simplifies the implementation of organisation-wide data strategies.

Cons:

• Speed is flagged as the number one challenge e.g. may lead to slower response times to department-specific needs.

• Risk of data professionals being detached from specific business contexts.

• Potential for bottlenecks due to centralised decision-making.

Decentralised Model:

Pros:

• Allows data professionals to develop deep domain expertise.

• Enables quicker decision-making within departments.

• Fosters a sense of ownership and accountability in data initiatives.

Cons:

• Can lead to inconsistencies in data practices across departments due to lack of knowledge sharing.

• Challenges in maintaining unified data governance.

• Risk of duplicated efforts and resources.

People and Fit

As the owner of a recruitment agency specialised in data & analytics folk you know what I'm going to say here:

It's not about which model is better, it's about the people in it.

Some professionals thrive in a centralised environment.

While others excel in decentralised.

What I'm getting at here is hiring for the business model and sharing that with prospective applicants is the most important. Know thy business.

Perth's Landscape

In Perth, organisational preferences vary:

• Mining and Energy Sectors: Often lean towards decentralised models to cater to the distinct needs of various operations and also the scale of operations.

Other sectors like banking, education, healthcare and government have mixed models often depending on leadership and size of the data team.

If there are only three data folk at a 300-person company, centralising makes a lot of sense!

As we see data maturity uplift as a whole, decentralisation may gain more popularity.

Hiring Considerations

When recruiting for data roles, clarity about the team's structure is paramount.

Candidates should be informed about whether they'll be part of a centralised team, embedded within a department or operating in a hybrid setup.

And applicants should know what they're getting into.

This transparency means alignment of expectations and helps attract professionals who are best suited to the organisation's model.

Hire for your model.

The reason for this article is because I wanted to learn more about data team models and I found the best way to do that is to write about it.

The best article to explain this and source for this article has by far been How to structure your data team by dbt. It offers valuable insights - highly recommend.

DR Analytics Recruitment

I'm Douglas - former data analyst and Founder of DR Analytics Recruitment. We're putting people first by building a community that connects top data and technology talent with the right companies. Our vision is an Australia empowered by data literacy.

Get in touch to learn more!

📧 Email: douglas@analyticsrecruitment.com.au

📞 Phone: +61 430 846 876

🌐 Website: https://www.analyticsrecruitment.com.au

⭕ Data Community: https://www.meetup.com/en-AU/industry-inner-join/