Data literacy, governance keys to transformation at Dow

Dow Chemical Company is one of the largest chemical producers in the world, with a presence in roughly 160 countries and more than 37,000 employees worldwide. But with analytics and AI becoming table-stakes to staying competitive in the modern business world, the Michigan-based company struggled to leverage its data.

“We have all these data scientists and business problems that would be more easily solved with data,” says Brandon Schroeder, the company’s IT director, data & analytics platforms. “We didn’t have a centralized place to do it and really didn’t do a great job governing our data. We focused a lot on keeping our data secure. We didn’t spend as much time making our data easy to use.”

Dow’s scale resulted in a large and diverse dataset that was isolated and cumbersome. It was difficult, for example, to combine manufacturing, commercial, and innovation data in analytics to generate insights. The lack of a corporate governance model meant that even if they could combine data, the reliability of it was questionable.

“We were looking at all the potential that was being left on the table, and the time to deliver insights for business problems was too long,” Schroeder says.

So in 2022, the company set out to change that with a project called the Integrated Data Hub. The idea was to dramatically improve data discoverability, accessibility, quality, and usability. But Dow didn’t just set out to create a centralized data repository. Its goal was to transform the way all its employees interacted with and related to data, empowering the entire organization to make data and analytics part of how they work. As a result, the project, which earned Dow a 2024 CIO Award for IT leadership and innovation, aims at four main outcomes:

• Better data access and literacy
• Data and analytics that anyone in the organization could use and benefit from
• An enhanced employee experience and ease of use by removing technical barriers
• A data culture and community that encourages transparency and reusability

“One of the softer promises we made with the Integrated Data Hub was to lower the barrier to entry into data and analytics, and data analytics literacy is a big part of that,” Schroeder says. He notes that Dow could put all the technology and data in place so 200 data scientists in the company could use it, “or we could train every person at every level of the company to take advantage of all this work we’ve done.”

Knowledge is power

Nathan Wilmot, Dow’s IT director, client partnerships, enterprise data & analytics, says the literacy program covers everything from teaching how to use gen AI and building data visualizations, to better managing data and making decisions with data.

“There’s a cultural change happening in Dow across data analytics and AI writ large,” he says. “What we’ve developed in parallel to the Integrated Data Hub is our data and analytics literacy program, which is really trying to drive the ability for everyone across the company, for their specific role, to utilize and communicate with data in the right context. If you’re a data generator or a consumer, analyzer, decision-maker — whatever it is, you have the right skill set to use the most modern technologies, applying data and analytics as efficiently as possible.”

The first step was to get buy-in for the project. With access to good, clean data essential to leveraging AI, Schroeder says it wasn’t hard to convince senior business leadership that the Integrated Data Hub was a necessary project. Still, he says, it’s hard to do anything at a company the size of Dow by yourself, so it was vital to seek partnerships across the company’s businesses, within the IT organization, and in Dow’s other functions. The security organization was an especially valuable partner, too.

“When you’re taking the whole of Dow’s 127 years of knowledge in the form of structured and unstructured data and putting it in a place that’s supposed to make it easier to access and find, that can be scary,” Schroeder says. “There are data privacy laws, and security regulations and controls that have to be put in place. Having a strong, deep partnership with our information systems security organization was a huge catalyst for us being able to go down this road. We had to rethink how we approach some of our cloud principles and how we approach data.”

A steep learning curve

Right out of the gate, though, the team made some mistakes. Schroeder says they tried to go too fast and solve too many hard problems up front.

“We were trying to skip over some of the data governance aspect with the idea that we would come back and go after that later,” he says. “We immediately saw the pains of that, and as we extrapolated what that would look like for the whole organization, we pulled back.”

That false start led to self-reflection.

“If you don’t know who owns the data, what that data is classified as, where it’s at, and you don’t get the buy-in from the people who truly own that data, all this other stuff is just a waste of time,” he adds.

“This is a process, and it takes time,” Wilmot says. “The more mature you are as an organization, the more challenging this is. But you have to do this to continue to be successful in the emerging world of data analytics, AI, generative AI, and all the things that will follow. The technology is evolving so quickly that you can’t not do this and still expect to be successful.”

A question of culture

With a renewed focus on data governance first, there were still major issues to overcome, especially the number of changes the project introduced to the organization and the extent of those changes. Schroeder’s team adopted a hub-and-spoke model for the Integrated Data Hub’s operational practices, requiring shared responsibility and ownership, modern technical skills, and a high degree of data literacy. Buy-in from business and functional leaders was essential as well.

“The challenge is mostly on the cultural side,” Schroeder says. “The technology stuff is easy. We’re out there asking for accountability on data as owners of that data, and sometimes that comes with extra tasks you have to do every day. You have to convince everyone why that’s a good idea and why it’s important you take on this extra work. Fundamentally, it’s about managing change.”

The team worked closely with experts across the organization to ensure the Integrated Data Hub would meet their long-term needs. That included choosing a technology platform that could integrate with downstream analytics capabilities being used across the organization, and developing a governance model that ensured proper data ownership and stewardship without creating an overwhelming burden on already stretched resources. The team also focused on business-aligned use cases to help prioritize the work, something Schroeder says ensured that the data entering the data hub was properly governed and provided real value.

The team coupled that work with a robust training and support model, including an Integrated Data Hub certification process that leverages a persona-based approach to provide individuals with training aligned with their work as data scientists, data engineers, data analysts, and data owners.

The team began loading data into the hub in 2023 and there was high demand for adding data products almost immediately. Wilmot notes that data accuracy has been climbing at an accelerated rate since the creation of the data hub.

“We’re seeing the accuracy consistently above 90%-plus, which is considerably higher than it was when we started,” Wilmot says.

In addition, Schroeder says time-to-insight has considerably improved. One of the goals he’s shooting for is something he calls “data science in a day,” in which a data scientist can get an idea and have access to all the data, platform, and environments they need to start working on the problem within 24 hours.

“We’re not there yet, but that’s the target,” he says.