Structuring the Analytics Team: Distributed vs. Centralized Approaches

In today’s data-driven world, the role of analytics in organizations is more critical than ever. As an analytics leader, structuring your team effectively can be the difference between merely collecting data and truly leveraging it to drive strategic decisions. There are two primary structures to consider: a distributed model, where each business unit (BU) has its own analytics team, and a centralized model, where a single, central analytics team serves the entire organization. Each approach has its strengths and weaknesses, and understanding these can help you determine which structure best suits your organization’s needs at different stages of its growth.

The Distributed Structure

In a distributed analytics structure, each business unit within the organization has its own dedicated analytics team. These teams work closely with their respective BUs, providing tailored insights and solutions that address specific needs.

The Centralized Analytics Structure

In a centralized analytics structure, a single, unified analytics team serves the entire organization. This team is typically housed within a central department and works on projects across different BUs.

Which Structure Works Better and When?

There is no one size fits all solution available when it comes to designing the structure of your analytics team, but generally, based on the pros and cons of each structure, there are general structures that could be followed depending on the size and complexity of the organization. However, these structures need to evolve with time as the organization and its needs evolve.

Early-Stage Companies: For early-stage companies, a centralized analytics structure often works best. These organizations typically have limited resources and a need for standardized processes to establish a strong data foundation. Centralization ensures that all analytics efforts are aligned with the company’s overarching goals and that resources are used efficiently.

Mid-Sized Organizations: As companies grow into mid-sized entities, they might benefit from a hybrid approach. This involves maintaining a central analytics team for overarching strategy, governance, and complex projects, while embedding a few analysts within key BUs to provide localized support and insights. This hybrid model combines the best of both worlds, offering the standardization and efficiency of centralization with the agility and contextual understanding of a distributed model.

Large Enterprises: Large enterprises with diverse and complex operations might lean towards a more distributed model. However, it’s crucial to implement strong governance and coordination mechanisms to mitigate the risks of resource duplication and lack of standardization. Establishing a central “Center of Excellence” (CoE) within a distributed structure can help achieve this balance. The CoE can provide strategic direction, standardized methodologies, and shared resources, while the distributed teams focus on delivering BU-specific insights.

Recommendations for Implementation

Assess Organizational Needs: Regularly assess the needs and maturity of your organization. Understand the specific requirements of different BUs and the overall strategic goals.
Develop Clear Governance Structures: Whether centralized or distributed, establish clear governance structures to ensure consistency, quality, and alignment with organizational goals.
Foster a Culture of Collaboration: Encourage collaboration and knowledge sharing across teams. Implement tools and practices that facilitate communication and the exchange of ideas.
Invest in Training and Development: Continuous training and development are crucial for keeping the analytics team updated with the latest tools, techniques, and industry best practices.
Monitor and Adapt: Regularly monitor the effectiveness of your chosen structure and be prepared to adapt as the organization evolves. Flexibility is key to staying responsive to changing business needs and technological advancements.

Razorpay’s analytics structure and its journey

At Razorpay, we’ve been through multiple stages of the analytics team’s evolution.

Stage 1: Initial Distributed Model

Razorpay’s analytics journey started with a decentralized structure, where analysts were dedicated to each business & product unit. During the initial phase, this enabled each BU to get their basic metrics & reporting up and running and to ensure that enough relevant data was available for decision-making.

Stage 2: Transition to Centralized Teams

We recognized the need for a more structured approach as the team grew. By 2019, the analytics team had grown to around 40 members, and the organization had expanded into multiple product lines. That’s when we transitioned to a centralized model to address the increasing complexity, forming two primary teams: business analytics and product analytics. The business analytics team focused on working closely with the business teams to enable data-driven decisions, while the product analytics team collaborated with product managers to scale products through efficient tracking of product metrics and conducting RCAs to identify the gaps.

Stage 3: Establishing the Center of Excellence (CoE)

With continued growth, reaching approximately 100 team members, Razorpay saw the need to leverage synergies between the business and product analytics teams. To achieve this, they established a Center of Excellence (Analytics CoE) in 2021–22. In this structure,

The product and business analytics teams were combined, and analysts were divided into pods. Each pod works closely with a specific business or product team, providing deep contextual insights.
There is a central group of analytics managers, with spans covering multiple pods (and often, the business & product parts of the same product)
The team takes on both vertical (product/business-focussed) goals, and horizontal (analytics excellence) goals and the manager group helps facilitate collaboration between different teams
Some examples of horizontal goals that we are driving: standardized processes for data instrumentation, product experimentation, optimizing data pipelines, standardized dashboarding practices, tool adoption & migration.
The central structure also ensures that analytics standards are established for things like: ways of working (sprint plans etc), role competencies & analyst growth paths etc.

Insights from Razorpay’s Journey

Razorpay’s journey illustrates the importance of evolving the analytics structure in response to organizational growth and complexity. By initially adopting a distributed model, we ensured close alignment with business and product unit needs. As the company grew, transitioning to centralized teams helped establish consistency and efficiency. Finally, the creation of the CoE allowed us to balance deep contextual knowledge with standardized processes, improving overall analytics quality.

Conclusion

Choosing the right structure for your analytics team is a critical decision that can significantly impact your organization’s ability to leverage data effectively. Both centralized and distributed structures have their own set of advantages and challenges. By understanding these and aligning your choice with the organization’s stage of growth and specific needs, you can build an analytics team that not only provides valuable insights but also drives strategic success. As your organization evolves, remain flexible and open to adapting your structure to ensure that your analytics capabilities continue to meet the demands of an increasingly data-driven world.