Organizational Design

The Data Product
Operating Model.

Technology alone cannot build a Data Mesh. You must design an organizational structure that aligns data ownership with business domains. Explore the three primary ways to organize your data teams.

Data Product Operating Models

The Evolution of Data Organization

As organizations mature and scale, their data operating models must evolve from monolithic bottlenecks to decentralized networks.

Option 1

Centralized

A single, central data engineering team is responsible for ingesting, transforming, and serving all data across the entire organization.

Pro: Easy to enforce governance and standardize tooling.

Con: Creates a massive bottleneck. The central team lacks domain context, leading to slow delivery and misaligned data.

Most Common
Option 2

Hub and Spoke

A central platform team (Hub) builds the self-serve infrastructure and standards. Embedded data teams within business units (Spokes) build the actual data products.

Pro: Balances speed and standardization. Domain teams move fast using central tools.

Con: Can be politically complex to manage funding and boundaries between Hub and Spokes.

Option 3

Full Data Mesh

A highly decentralized model. Cross-functional domain teams completely own their data products, infrastructure, and pipelines, communicating strictly via standardized ports.

Pro: Infinite scalability. No central bottlenecks. Maximum domain autonomy.

Con: Extremely high engineering maturity required. Risk of duplicated effort and siloed infrastructure.

The Path Forward

Choosing Your Path

There is no "perfect" operating model. The right choice depends entirely on your organization's size, engineering maturity, and the complexity of your business domains.

Most enterprises successfully adopting the Data Product mindset currently operate in a Hub and Spoke model. It provides the necessary autonomy for domains to act like product teams, while relying on a centralized platform team to enforce DaaP standards (Discoverability, Security, etc.).

Decision Matrix

When to use Centralized:

  • Small startups (< 200 employees)
  • Single, simple business domain

When to use Hub & Spoke:

  • Medium-to-Large enterprises
  • Need speed but lack high engineering maturity in every domain

When to use Full Mesh:

  • Massive, global tech conglomerates
  • High software engineering maturity everywhere

Restructure for Scale

Identify where your organizational bottlenecks are today, and map out your transition to a decentralized, domain-driven data structure.

Back to Mindset