Designing for scale – building a data team structure that delivers for humans and AI

Growing data volumes mean that organizations have invested heavily in technologies that help them manage it, ensuring compliance and good governance. Now, Chief Data Officers are looking to make this data available, at scale, to the entire business and AI agents. 

Traditional, centralized data team structures struggle to cope with this transformation. Central data analysts are overwhelmed by a growing number of requests for data and reports, while valuable data remains unused because it cannot be either found easily or its value is difficult to understand. Technical data and IT teams don’t have the specific business knowledge to focus on the data that users actually need. 

All of this holds back the impact of data and prevents organizational agility. How can companies balance central control with ensuring that business needs are met?

Drawing on the latest Gartner research, as shared at the Gartner Data & Analytics Summit 2026 London, this blog distils the latest best practices and options, explaining how they can be deployed to create structures that scale data consumption. It highlights the importance of self-service data product marketplaces to successful data and analytics (D&A) organizations, outlining how they provide a single, collaborative space for data across the business by bringing together data, business, IT and AI teams.

There have traditionally been two main approaches to managing data and making it available to the business:

• A global, centralized hub team that leads and controls data sharing, ensuring consistency
• Decentralized regional/domain (department or office)-specific spoke teams that bring context and business expertise to data

However, in a complex and fast-moving world, simply choosing one of these two options is not enough. Instead, Gartner recommends studying five key best practices and applying them to your D&A team structure and operations, based on the specific circumstances and needs of the business.

1 Look at the franchise model

In the business world, franchises provide a combination of central control from a brand owner, while encouraging entrepreneurship and agility based on local understanding. They enable companies, particularly in areas such as hospitality, to scale quickly while still delivering a consistent experience across every location.

Analytics franchising follows the same model. Just as in a fast-food restaurant chain, decentralized spokes (such as in business units or subsidiaries) operate independently to meet local needs, but within frameworks and guidelines set out and enforced by a central team. This ensures consistency and governance, avoids reinvention and overlap, and enables flexibility and agility by empowering decentralized teams.

Franchises also offer the ability to be rolled out on a gradual, bottom-up basis, rather than being implemented everywhere at once. Chief Data Officers can identify the departments and teams best suited to franchising through their maturity, tech and analytics skills, and then work with them to create the local organization. 

To drive success, Gartner has six key recommendations:

• Identify and recruit business leaders to become owners/operators of an analytics franchise
• Create a balanced organizational model with a centralized team working collaboratively with decentralized analytics teams
• Assemble cross-functional teams that blend IT skills, analytic expertise and domain knowledge
• Establish analytic sandboxes where small cross-functional teams can quickly integrate data and create analytic prototypes
• Provide each analytics franchise with the tailored ability to create prototype, pilot and production content, depending on maturity and skills
• Establish strong lines of communication to promote analytic content between analytics franchises and centralized teams

2 Adopt federated data management

A federated approach splits different responsibilities between local and central organizations to ensure consistency and agility. Key areas around compliance (such as governance) are agreed and enforced on a company-wide basis, while ensuring scalability and an approach that meets local needs. 

“Organizations that decentralize data management teams and follow a federated data governance model are 3 times more likely to deliver valuable business outcomes compared to those that do not.” Mark Beyer, Gartner

While the structure of every business is different, based on talking to CDOs across the globe, Gartner has created recommendations for seven areas that help guide how data teams are organized.

3. Fusion teams

Gartner defines fusion teams as “a multidisciplinary team that blends information technology or analytics and business domain expertise and shares accountability for business and technology outcomes.”

Essentially, instead of roles being assigned either to local business teams or central data/technology departments, fusion teams are organized around specific business capabilities or outcomes. For example, they could focus on using data to solve a problem such as sales churn or improving the customer experience.

Fusion teams bring together:

• Data personas (such as data engineers, DataOps and data stewards)
• Technology personas (for example, software engineers and DevOps experts)
• Business personas (business analysts, data product owners and end-user representatives)

By including a data product owner, fusion teams are able to build, deliver, monitor and maintain high-impact, governed and easily consumable data assets that are designed around specific business needs. This helps scale data consumption by both humans and AI.

4. Design the model from the ground up

Traditional data management organizations have grown organically, adding new responsibilities and technologies over time. While it is impossible for businesses to simply demolish these structures and start again, thoroughly reviewing the current organization against requirements provides a plan for change over time.

Gartner recommends starting by asking the question “What does your overall organization need?”, and then using this to create a structure that delivers strategic impact, scales data use, increases efficiency, and has the right technology and skills in place. Depending on requirements, the model should accommodate different roles, such as:

• The advisor – advising senior management on how D&A can enable business impact
• The pollinator – building awareness, understanding and adoption of data at scale 
• The conductor – co-ordinating efforts across the organization and sharing best practices 
• The architect – designing and creating technical processes and standards
• The pathfinder – creating innovative, early stage proof of concepts
• The engineer – developing and operationalizing projects when they move beyond proof of concept stage
• The provider – managing and providing platforms to underpin data sharing and consumption
• The agency – building and providing the right skills and talent across the organization
  

5. Think AI-first 

Adopting the right structural and operational framework to organize AI initiatives is vital to scaling deployments and benefits. The AI organization therefore includes the development of internal teams, the establishment of external partnerships and the strategic alignment of Al efforts with business objectives. It links closely to data teams, particularly if both are the responsibility of a Chief Data & AI Officer (CDAO). 

Given the relative immaturity of AI initiatives in many businesses, and the rapid pace of change, the exact balance between centralized and decentralized capabilities is likely to shift over time. As capabilities become more mature and dispersed, structures should become more decentralized to ensure AI initiatives match specific business needs, while still meeting corporate guidelines and best practices. Gartner offers five recommendations to create an AI organization that evolves with your needs:

• Start small and iterate
• Establish leadership structures
• Invest in platform capabilities early
• Foster communities of practice
• Continuously monitor and adapt

Putting in place an intuitive, self-service data product marketplace connects business users to the data they need. Importantly, it also supports CDOs in optimizing their data team structures in six key ways, underpinning the best practices highlighted by Gartner:

Providing a central collaborative space

Data product marketplaces bring together local data product owners and central data governance and administration teams in a single place. They can collaborate around creating, sharing and promoting data products, including monitoring their uptake and use cases. They support fusion teams and analytics franchises, bringing together employees from the business with technical experts to create a community around data.  

Enforcing governance and best practices

All data assets published on the data product marketplace are checked to ensure they meet governance, data quality, and security standards, whoever creates them within federated data management models. By applying processors, data assets are standardized around corporate guidelines, both around technical formats (such as dates) and reference data, as well as business terms through business glossaries. As a data product marketplace provides a single version of the truth, the organization benefits from consistency and auditable data processes.

Enabling seamless data sharing by all

Data product owners from across the business can easily create and share data assets through the data marketplace in a straightforward way, making them available to users. Access can be controlled to sensitive information if required to ensure security and confidentiality while maximizing reuse at a local, domain and/or global level.

Delivering data for AI innovation

Successful, scalable AI requires access to trusted and contextualized data.  Data product marketplaces make data assets, especially data products, available in machine-readable formats, enabling them to be seamlessly used by AI models and agents. Usage information helps deliver context to ensure accuracy, while lineage provides monitoring and transparency. Built-in data contracts set out how data can and can’t be used, minimizing risk and maximizing innovation.

Showcasing data reuses

Data from one part of the business can be used in completely new ways by other departments or teams. Data product marketplaces underpin and encourage this sharing, both by making data available in understandable, usable formats and also allowing users to share their reuses with the wider organization. This spurs new ideas and uses, drives collaboration, and enables innovation.

Building a company-wide data culture

Collaboration and communication are essential to building and spreading a corporate data culture. By making data easily accessible to all users and data teams, data product marketplaces build trust in data and encourage everyone to experiment, share ideas and provide feedback. This helps create the skills and culture required to scale data consumption within the organization.

As organizations look to increase data sharing and use, they need to have the right structures in place, balancing centralization and governance with local, business requirements. An effective data product marketplace supports these structures, helping organizations optimize data sharing, embrace innovation and drive data democratization.

Want to learn more about structuring your data teams? We’ve worked on over 3,000 data marketplace projects across the globe – talk to our experts to find out how we can help you deliver on your objectives.