6-minute read

Quick summary: Utilities can build organizational capabilities in data science with a focus on technology alignment, strategic innovation, and talent management in a center of excellence.

Today’s utility companies face an unprecedented array of challenges, from an aging infrastructure to increased electrification to the rise of distributed energy resources (DERs). Successfully addressing these challenges requires a broad range of diverse solutions that have one vital element in common: they all rely on strategic use of source data, which in turn requires strong capabilities in data science.

Just as the need for data analytics in the energy sector is not limited to one or two departments, building organizational capabilities in this discipline requires an approach that touches on all areas of the business. When we work with utilities on data science, we take a hybrid team approach, working side-by-side with their in-house experts to build their organizational capabilities through a data science center of excellence (CoE).

What is a data science center of excellence for utilities?

A center of excellence is simply a single team that focuses the vision, strategy, and infrastructure for a discipline—a particularly useful approach for data science due to the highly specialized skills it requires. The CoE team typically acts as an internal consultancy, working with divisions across the organization to identify and capitalize on opportunities to apply data science in developing data-driven solutions.

While the center of excellence is not the only model for developing organizational capabilities, it has delivered successful outcomes for us in a large number of client engagements. Through our hybrid team approach, we’re able to walk our clients through the building and implementation of their own CoE capable of driving the organization towards its data science goals.

Why establish a utilities data science center of excellence?

In the absence of a center of excellence, each department is tasked with hiring its own data scientists and establishing its own standards and processes, which creates significant inefficiencies and hinders standardization. In addition, each team would work as a silo, building hyper-specialized tools that are difficult to integrate and maintain over the long term.

A center of excellence offers the advantages of cost centralization and the flexibility needed to bring data science solutioning to different areas of the business, wherever and whenever a need arises. So even areas with minimal or infrequent data science needs, such as human resources, can gain access to data science capabilities without having to staff up on their own.

I like to use the metaphor of a community fire department. Sure, every property owner could maintain their own firefighting personnel and equipment, but it’s far more efficient and cost-effective to have a centralized unit that serves everyone.

Three pillars of building utilities’ data science capabilities

Integrating data science in the energy sector is not simply a matter of hiring some data scientists and putting them to work. Building effective data science capabilities requires a coordinated series of initiatives focused on technology, strategy, and talent, all of which can be guided by the CoE team.

Technology alignment

When most people think about building utilities’ data science capabilities, the technological aspect is usually the first to come to mind. Addressing this aspect requires first aligning technology solutions with utilities’ specific business needs and future use cases for analytics—answering the question “Which problem(s) will this technology solve?” and configuring systems accordingly. When technology alignment is successful, the utility will benefit from a stack of industry-best machine learning tools, solid and maintainable deployments, and a scalable machine learning operations (MLOps) infrastructure for solving enterprise problems.

A utility’s data science CoE team can guide the organization through specific technology alignment activities, which might include

• Building a standardized tech stack for machine learning applications

• Creating a CI/CD pipeline for models

• Designing a scalable data and application architecture for batch, real-time, and interactive use cases

• Automating the model lifecycle

Strategic innovation

Delivering business value through data science requires a strategic approach to the management and governance of data science projects. A utility’s data science CoE can lead the way in developing a strategy and processes for realizing value and team success.

When the strategy and process side of building data science capabilities is successfully addressed, the utility will have a vision for delivering value to the business, a system for adoption and prioritization planning, and an ecosystem for iterative development and delivery of data science solutions.

Strategic innovation may encompass activities such as

• Defining an Agile data science working process

• Identifying and prioritizing a backlog of opportunities

• Instituting a model governance process to ensure quality, security, value, and maintenance

• Creating and evangelizing an opportunity intake process

Talent management

We find that the talent aspect of building organizational capabilities is the most often overlooked, but it is no less important—possibly more so—than the other two pillars. Utilities must set a vision for their data science capabilities that leverages and develops the strengths of their teams.

Successfully retraining, retooling, and reskilling a workforce requires utmost attention to the human elements, and that’s where the practice of change management comes in. By leveraging proven change management strategies, utilities can ensure that employees not only have the information they need, but also understand their roles in achieving the goals of the initiative and how it benefits them personally and professionally. When a utility successfully addresses the talent management aspect of data science capabilities, it will have skilled and domain-knowledgeable teams, training and goals that build up the capability, and a set of clear expectations, communications, and practices.

Specific activities focused on the talent aspect of building data science capabilities may include

• Assessing existing talent and setting career paths

• Clearly defining data scientist and machine learning engineering roles

• Creating training pathways for data scientists and engineers

• Creating internal cross-training initiatives

Building a data science–driven utility

For utilities looking to address the challenges of today and tomorrow, data holds the key, and developing organizational data science capabilities lays the foundation for future successes. By devoting time, attention, and resources to the technology, strategy, and talent aspects of this organizational shift, utility companies can build a strong culture in which data achieves its full potential as a strategic asset.

Adam Cornille is Director of Advanced Analytics at Logic20/20. He is a data science manager and practitioner with over a decade of field experience, and has trained in development, statistics, and management practices. Adam currently heads the development of data science solutions and strategies for improving business maturity in the application of data.