Empowering a Data Culture From the Inside Out

For companies that struggle with data transformations, underthinking organizational change is often a bigger problem than technology issues. A company can have powerful tools and meaningful data at its disposal, but without the proper education and processes to put that data in the hands of the right people and provide business context, extracting value can prove difficult.

In 2016, Jonathan Tudor founded a self-service data program at GE Aviation aimed exactly at this problem. By recognizing that success would depend on empowering users beyond the data engineering and analytics teams, he was able to encourage buy-in from across the organization, increase engagement, and create cross-functional partnerships.

Ally MacDonald, senior editor at MIT Sloan Management Review, spoke with Tudor about his work with self-service data and organizational transformation. What follows is an edited and condensed version of their conversation.

MIT Sloan Management Review: What is a self-service data program?

Jonathan Tudor: The idea with self-service data is, rather than hiring endless numbers of highly competitive data talent, why not take your existing intellectual capital and people capital within the company and empower them to do their own data analytics work? In a self-service system, line-of-business professionals and analysts in the company can access and work with data and data visualization directly, and they are supported by, but not dependent on, IT and data professionals to carry out their work.

This kind of program allows companies to remove technical boundaries and empowers people to use their own subject matter expertise — after all, they know the problems they’re tackling best, and they know what data they need — to generate insights and execute their work.

When data is fundamental to how you run your organization, analytics are needed in all parts of the company very quickly and are key to business outcomes.

How has this new approach evolved to play a role in organizations?

Tudor: It’s helpful to look back in history. When we think of business intelligence [BI] and data warehousing, this has often been siloed within organizations. BI teams do all the work to gather the data, structure it (and hope it’s structured correctly), and then deliver those insights to the customer. That process is similar to what we see in a waterfall method in IT, which is very sequential and dependent on the people carrying out each task.

But as companies push ahead with more data than ever and are utilizing things like data lakes, a waterfall approach becomes less effective. When data is fundamental to how you run your organization, analytics are needed in all parts of the company very quickly and are key to business outcomes. The truth of the demands for many organizations is that they will never be able to hire enough data engineers or data scientists to meet these growing business needs on their own.

Self-service data allows companies to meet these demands by focusing less on who is carrying out the work, in favor of the business outcomes.

What KPIs are important when it comes to measuring the effectiveness of self-service data?

Tudor: Companies will vary in their approach and certain metrics will be more important depending on the stakeholder. At GE Aviation, we focus on three self-service data metrics that shed light on financial benefits, how much value users are getting from the program, and innovation viability for the business.

First, there are financial KPIs. We can determine where there are impacts on the balance sheet income statement or how much revenue we can track from a return perspective.

The second important metric is utilization. We look at the number of unique users and track what percentage of them are staying active in a given period. For example, we might find that the self-service program has achieved 2,000 unique users across the ecosystem in the last three months.

A third area involves creating an innovation pipeline from tracking engagement and usage. For example, we look at every data artifact that has over 50 unique users in a month — every BI report, every analytic. Based off that, we can inform leadership about what is proving most important to users and may be worth further business investment. This helps us to create an organized pipeline that shows the value proposition of new, innovative ideas within the organization.

If self-service systems are key for enabling faster decisions in the organization, that means many teams and units throughout the business are now becoming more hands-on with the data. What does that mean for data governance?

Tudor: Self-service data and governance definitely go hand in hand. What I liken it to is, you’re trying to run a playground: You want to empower people so that they can do what they need to do and do it well, but you also want to keep them from hurting themselves or others.

There’s a delicate balance here from the start. You need to protect users and the company, ensure that you’re compliant and meeting regulations, and enable better understanding of data. But at the same time, you don’t want to be such a barrier that people can’t do their work.

I’m a strong believer in building self-service programs from the ground up with governance and an understanding that they will need to scale. Things like data cataloging, data lineage, and providing the business context around the data are all very important parts of governance, because people need to have the right information and context about the data they’re looking at in order to be successful.

What are other challenging areas that companies should be mindful of when approaching these types of initiatives?

Tudor: Data stewardship is also very important. Today there are lots of ways for companies to automate their data catalog using software to help make decisions. But ultimately, you still need to have people in the company acting as data stewards — that is, providing business context and serving as points of contact for others in the organization who may have questions or need more information. This is often very difficult to achieve in companies, because being a data steward is not commonly a dedicated role, and it’s challenging to get people to commit their time to something that is not necessarily their primary job.

Something we learned early on to help incentivize data stewardship involved implementing gamification. We started tracking activity of users in our ecosystem and created a related point system and leaderboard that ranked people using the self-service program. People could then compete against one another and be recognized monthly for their contributions.

I’ve seen many organizations struggle with this, and I will admit this is still hard today. But by making it more interactive and even fun, we’ve significantly improved participation.

What should data-focused managers be thinking about when it comes to onboarding talent today? What do companies and managers overlook?

Tudor: The talent you’re looking for often needs to be different than [what] we generally think about when we talk about IT or data and analytics. You can’t just be a technologist today. Oftentimes, the problem you’re going to encounter won’t be a technology problem but a people problem.

On my team, we often think of ourselves as a startup within a larger organization, where you need to wear many different hats. We have many people who have nontraditional backgrounds for IT — for instance, former musicians and chefs who switched careers and bring diverse experience to their work in data and technology. I think this diversity helps in many ways for connecting with people, because much of our work is people-oriented in addition to being driven by good technology.

The other thing that’s really important for training talent is stressing the importance of the business partnership that needs to take shape for any self-service program to be successful. For example, we train and partner with dozens of data ambassadors — professionals not on our central data team but who can act as extensions of the team in locations across the globe. They help manage communication, break down organizational barriers, and drive governance across different sites. Encouraging this type of collaboration is critical for amplifying the role of a smaller self-service team across the wider organization.

It’s clear from our discussion that these systems aren’t static — teams are continuously measuring and learning. How do managers shift their thinking and way of managing traditional business intelligence to support self-service data success?

Tudor: One thing for managers is simply being OK with data and analytics or IT not being done by teams who are in data and analytics or IT. That’s a very fundamental one, but it can be very hard. You have to be comfortable with letting go of some control.

In some ways, it’s like you’re an app store and you’re helping a self-service community build apps that might not be developed on your team. This means becoming more of a systems thinker in order to drive a design that empowers other people to be successful at scale. This is likely the biggest and most difficult mind shift to make, but there’s a lot to gain from embracing it.