Background Circle Background Circle

Data Mastery in the AI Era: Dr. Tom Redman’s Blueprint for Excellence

During his 15 years at AT&T Bell Labs, Redman founded the first data quality lab of its kind and developed the underpinnings for data quality management. Three decades later, he runs a global business spanning financial services, defense, tech, telecom, consumer goods, and more.


DATA LEADERS NETWORK: For those unfamiliar with your work in the data space, briefly describe what you’re working on and how you envision the space in 1-3 years. 

REDMAN: The most important thing I’m working on today is “regular people,” which is the label I give those without data in their titles. There are enormous opportunities for them to step into the data space, empower themselves, and fix something. When they do, it enriches both their personal and professional lives. Which is why I do everything I can to convince companies to bring as many regular people as they can into their data — especially data quality– programs. Without high-quality data, the potential to do real harm is enormous. 

I also work on data quality for AI. Right now, the data space is dominated by AI. The hype is enormous, and everyone is all over the map. A few are getting traction, but most are not. Many are frustrated, because it takes a lot to succeed with AI (see New Technologies Arrive in Clusters. What Does That Mean for AI? Those who can see through hype and do some foundational work will have a real advantage. They have to address data quality properly, get everyone involved, and build some other organizational capabilities. 


DATA LEADERS NETWORK: In your Harvard Business Review article What Does It Actually Take to Build a Data-Driven Culture?, you assert that bad data is the norm. Why is this the case and what are the most accessible solutions that businesses can implement with minimal financial investment to create an environment where good data becomes ubiquitous? 

REDMAN: Pick practically any department. For example, Sales. You’ll find that they spend a significant fraction of their days correcting errors in the leads data they receive from Marketing. It’s time-consuming, frustrating, expensive work — and not usually done very well. The better approach would be for Marketing to make fewer errors. But it’s not like Marketing comes into work saying, “let’s send bad data to Sales.” They simply don’t know. Sales has to explain their data needs to Marketing, both need measurements in place, and Marketing must find and eliminate root causes of error. It’s pretty basic stuff. 

Across the company, the secret is helping people to recognize that they’re both “data creators” and “data customers.” In the example above, Marketing is the creator and Sales is the customer. When people step into these roles, quality improves quickly. 


DATA LEADERS NETWORK: For brand, product, or creative-focused companies (i.e., Nike), data integrity is usually very low on their list of investment priorities. What are some of the key points you’ve used to convince companies that investing in high-quality, usable data should be a priority? 

REDMAN: Not to make too fine a point of it, but companies are already investing in data quality. The problem is that they don’t get much for their money. The good folks at McKinsey estimate that 30% of people’s time, across the company, is spent on non-value-added work associated with data. So the issue is less about sheer money and more about spending it wisely. I don’t have any simple prescription for getting companies to see this. Change, no matter how small, is hard. Data quality management changes are pretty significant. 

That said, I do find two prerequisites: a gnawing business problem and an open mind to address it differently. For example, someone at any level in an organization grows dissatisfied with how their team, department, division, or company is handling a business issue and has an open mind to change. Some issues include, “I don’t trust these market share numbers,” “we’re having trouble with regulatory reporting,” or “people complain about the data at every staff meeting.”

The open minded cast a wide net to find a better way. They import the approaches, techniques, and management changes needed to make improvements. Often that involves data quality. These people are so important, I’ve given them the name “provocateurs.” In succeeding, they not only resolve their problem, they provide a script for others to follow. Others say, “I want that!” or senior managers say “I want that for everyone” and build a team tasked with leading the effort. Sometimes this doesn’t happen, of course. But when it does, it’s thrilling. All this goes faster the more senior the provocateur. The higher the better!


DATA LEADERS NETWORK: In your MIT Sloan Management Review article What’s Holding Your Data Program Back?, you discuss five key areas in the data space, and define the restraining and driving forces for each. Taking a more macro view, what restraining forces are responsible for keeping orgs inundated with bad data, and what driving forces are helpful in creating good data? 

REDMAN: The restraining forces called out in that article are the most common ones, but each company will have them in its own special ways. It’s like Tolstoy said, “All happy families are alike; each unhappy family is unhappy in its own way.”

As for those with good data, the first two things I look for are: the level of the most senior person demanding high-quality data, and the number of people who own their roles as “data creators” and “data customers.”


DATA LEADERS NETWORK: Is there anything else you’d like to share on the topic of data? 

REDMAN: Culture change is won through deeds, not words. 

It takes a lot of courage to make the changes that need to be made. Up and down the organization. But it’s possible. Make it clear that data and tech are different sorts of assets. Co-mingling the two almost certainly means you’re managing both poorly. Then, engage both the head and the heart. Engage the head with a business case and the heart with how the data program will advance things they value. Finally, break up the work into a sequence of simple, doable steps. No miracles required. 

Join the conversation on LinkedIn.