Read - Organizing for Success, Part I: How to Organize a BI Team in the Age of Self Service

The first article in this series discussed the role of a business intelligence (BI) team in the age of self-service. It argued that a strong BI team accelerates self-service, rather than impedes it. This article moves beyond a single team and addresses a question we often get asked: how do you organize a data analytics program to maximize value for the organization? Although there is no right or wrong way to do this, several patterns emerge when you examine successful organizations. 

The Federated Organization

A data analytics program spans the enterprise. It has corporate teams (e.g., the BI team, data management team, and data science team) that are sometimes called “centers of excellence” or “competency centers” because they provide coaching and support to people in the field. And it has business unit teams that consist of departmental data analysts, report analysts, data scientists, and analytics managers in sales, marketing, finance, etc. who generate insights and build reports for their colleagues. Thus, a data analytics program is federated—at least when corporate and business unit teams are aligned, which often is not the case. 

A data analytics program also spans levels. At the executive level, it has a “Data Analytics Council” that oversees the program and functions as a board of directors, and at the worker level, it has a “Community of Practice” that enables data analysts from business units to share ideas and techniques. It also has a partnership with IT systems owners and their database administrators that exists to alert the data analytics teams to any changes in source systems. (See figure 1.)

Figure 1 – A Federated Data Analytics Program

The Role of Standards 

The goal of a federated data analytics program is to gain the benefits of both centralization and decentralization without the downsides of either. Central teams set standards that ensure data consistency and economies of scale across the enterprise, but create bottlenecks and impede business agility. In contrast, business units can build good solutions quickly, but create data silos that can’t scale. In contrast, a federated organization provides both agility and scale, flexibility and consistency.

The goal of a federated data analytics program is to gain the benefits of both centralization and decentralization without the downsides of either.

Standards are the glue that holds a federated organization together. Standards include everything from tools, platforms, and infrastructure; to KPIs, data definitions, and governance processes; to self-service policies, project prioritization techniques, and visualization standards. Standards facilitate reuse, accelerate adoption, and ensure data consistency. Without standards, an organization devolves into data fiefdoms, which increases overhead costs and fragments data, creating data chaos. 

As one data analytics leader says, “Just as a car needs brakes to go fast, a data analytics program needs standards to go fast.” 

Typically, the corporate teams (BI, data, and analytics) define standards in conjunction with representatives from business units. The corporate teams then present the standards to the Data Analytics Council, which reviews, refines, ratifies, and then relays those standards back to the business units. This keeps a healthy balance of power between corporate and business unit interests and fosters communication and dialogue among all parties. This is critical to maintaining alignment and delivering long-lasting business value. 

Business Engagement Models 

At the heart of a federated data analytics program is a business engagement model—or, how corporate BI, data, and analytics teams interact with their business counterparts when building solutions that the business wants. 

Traditionally, companies hire an army of business requirements analysts (a.k.a. business analysts or BI analysts) who serve as translators between the business and corporate developers. This approach often doesn’t work well with data analytics because most business users don’t know what they want until they see it. Some companies supplement business analysts with “relationship managers”— senior people versed in both business and technology who form a strategic partnership with the business and proactively suggest solutions, rather than just take orders. 

Consequently, most large companies today have embraced agile methodologies with self-organizing teams, which consist of a data architect, BI developer, and data engineer, along with a business representative who prioritizes requirements and reviews output every two- or three-week cycle. (See Best Practice #6 in “Best Practices in DataOps: How to Create Robust, Automated Data Pipelines”.) Some companies preface agile development with a monthly or quarterly retreat during which business unit representatives present their requirements for the following period and collectively prioritize projects based on the known capacity of agile teams. 

Summary 

Companies that spend millions of dollars on data analytics should consider implementing a data analytics program to manage these investments. A data analytics program governs the people, processes, and technologies that an organization uses to drive insights and value from data. 

The best data analytics programs are federated: that is, they have corporate data analytics resources (a.k.a. centers of excellence) who support data analytics professionals in the business. The keys to a successful federated data analytics program are universally applied standards and a flexible business engagement model that supports iterative development with high-levels of business involvement. Both align corporate and local data analytics resources into a coherent, high-performing whole. 

The next blog in this series will drill into more detail about the roles and responsibilities of the various teams and committees within a data analytics program.

Read - Organizing for Success Part III: How to Organize and Staff Data Analytics Teams