Data Mastering in Conglomerates: A Korean Adventure

For chief data officers (CDOs), deploying master data management (MDM) is a necessary pain.

But wading through the host of challenges, ranging from data silos, incompatible systems, poorly defined use cases to interdepartmental politics, is not for the faint-hearted. These are also reasons why many MDM projects fail.

In conglomerates, the myriad of businesses and independent data infrastructures compounds the challenge. It was what Young Kim, group chief data officer at SK Group, the third-largest conglomerate in South Korea, faced when he was tasked to drive a groupwide MDM strategy.

Different starting points

Think of a conglomerate as a mini ecosystem of global industries. Each business is run independently and collates data separately from the rest. This compounds the MDM challenges as every company poses unique data challenges.

To create an MDM strategy for SK Group’s 95 subsidiary businesses, Kim had to ask the right questions that mattered to each business.

“Each company has different maturity levels when it comes to business and working together. Asking the right questions and understanding how we start this particular journey was one key portion. The other portion was knowing what they have and what they don't have and how do they put that information together. This was certainly the starting point.”

For example, in SK Construction, Kim’s team focused on ground workers at construction sites. They started an MDM project around the applications these workers used and showed how worker efficiency could improve. It created a strong use case to scale the project companywide.

Instead of getting all companies to a baseline data fitness for MDM, Kim and his team focused on the leaders like SK Hynix, the world’s second-largest memory semiconductor manufacturer.

Kim observes that SK Hynix had already “done much of the groundwork.” It created a champion within the group for other businesses, especially the smaller ones, to follow. It also created “silent pressure” for the laggards to follow.

Still, Kim believes he has his work cut out. Since he and his team is not part of any particular company, it is difficult to manage the internal people expectations and interdepartmental politics that often come with data sharing. “You have to have that conviction that what you're talking about will allow them to do a better business.”

Technology is part of the solution

Kim attributes his current success to the choice of technology solutions. One of them, Tamr, helped the companies to get on the MDM bandwagon. “What Tamr has done is completely transformed how an organization should look at creating an MDM program. And the reason why I say that is because they made it easy, fast, and more inclusive,” says Kim.

Inclusiveness is a significant factor in getting buy-in for MDM. It allowed all the business users, not just the data science teams, to drive change from within. 

Creating a center of excellence (CoE) for data from the onset helped. Kim called it his transformation team. “We would actually send out a group of people to work with the different companies,” says Kim.

To drive adoption among smaller companies, SK Group launched Multiverse with Google Cloud. As an integrated digital platform developed by SK Holdings C&C with the hyperscaler, it will combine major platforms and solutions for AI, big data, and the cloud. Smaller companies can log into the SaaS platform to use the data to create new use cases.

“That's where we are deploying Tamr along with Snowflake to drive a more modernized database environment,” says Kim.

Small wins are emerging 

When it comes to talent, there are no true and tried best practices. And for Kim, it is not the technical data knowledge that he sees as lacking, but those with business application knowledge. “So, data scientists may look at the data but would not care about the business perspective because they think it’s not their problem. And there a lot of people like that.”

Another challenge is data trust. For example, a sales leader will not care about MDM if they already have relationships with the top customers or have other means to drive his remuneration. So, Kim had to double down to convince these users.

As a result, his team experienced “a number of stops, starts, and restarts.” But he admits that no one ever thought MDM was going to be easy. Instead, it was more important to get a strong “consensus around what you are trying to solve really” to keep the project on track.

This conviction is finally bearing fruit. Different companies are beginning to see the value in using data to drive their business. For example, the SK Group’s energy business uses customer behavior data from their gas stations (a separate company) to identify what triggers their behavior.

Kim is confident that more such use cases for data efficiency will emerge as MDM becomes a foundation for all companies. Regular webinars, both internal, groupwide, and with external teams, also raise MDM awareness to new levels.

In the future, AI will help to do more of the heavy lifting of data. “I think Tamr already leverages AI very well. And I also foresee that these data scientists will leverage deep learning and other machine learning environments to help to speed up things and automate things in the near future,” Kim observes. 

Having a groupwide MDM strategy is not a means to an end. Tamr’s agile and quick time-to-value approach using human-guided machine learning shortens the time to produce results from years (with traditional MDM) to weeks. It helps companies to quickly unlock the value of data and show quantifiable ROI to the business, speeding up their data transformation initiatives.  

In fact, Kim views MDM as the start of SK Group’s data journey. The next step is asking the right questions using the right set of data sets. This, he feels, can help the companies and the group to recalibrate their business models quickly.

“But that is just our next step. For now, our number one priority is establishing MDM. And it is still a work in progress,” he concludes.

Winston Thomas is the editor-in-chief of CDOTrends, HR&DigitalTrends and DataOpsTrends. He is always curious about all things digital, including new digital business models, the widening impact of AI/ML, unproven singularity theories, proven data science success stories, lurking cybersecurity dangers, and reimagining the digital experience. You can reach him at [email protected].

Image credit: iStockphoto/NeoPhoto