Simon Hay is CEO of dunnhumby, a wholly-owned subsidiary of U.K. supermarket chain Tesco, that provides analytic services to retailers and consumer brands. Dunnhumby describes itself as a “customer science company” because it gathers and applies vast amounts of data to interpret and understand customer preferences and behaviors.
Founded 25 years ago, dunnhumby is a pioneer in big data, which of course represents a huge opportunity for businesses in all industries—as evidenced by Tesco’s investment in and ultimate acquisition of dunnhumby. Hay, who joined dunnhumby in 1991 and was appointed CEO three years ago, was named a Top 10 CEO in the U.K. earlier this year by Glassdoor, an online community.
What follows are highlights of a conversation with Hay about changes in technology, the evolution of the science of analytics, and the potential pitfalls of data overreach.
Dunnhumby has been around since 1989, which is a long time for a technology company. How has it evolved over time?
Simon Hay: In a way, the history of dunnhumby is inextricably linked with the history and growth of technology. When the company was set up, the premise was information. At that time, it was locked up in legacy systems. We believed we could extract and analyze and make some sense of data in a way that was more insightful than traditional methods of understanding customers, which, at that time, was market research—as in talking to customers and asking some questions, or some techniques like geodemographics, looking at where your customers lived and using the statistics about the area to describe the customers. So as dunnhumby has evolved, we've always pushed technology hard to get the best out of it and the most out of it to deliver more insight to us and our clients.
One insight seems to have been that looking at what people actually do is more telling or maybe a better predictor of what they will do than asking them what they will do. Where did that insight come from?
Hay: If you look at what people claim versus what people do, it’s no more apparent than when you talk to people about health. [Based on what people claim], we are all a maximum of 150 pounds, exercising five days a week, eating very low fat, and never see a physician. Of course, the reality of health data, people’s consumption of doughnuts, for example, is somewhat different.
So you have claimed behavior and you have actual behavior, and they both have a role, but the power we developed over time is the ability to put the two together. Sometimes, research is about aspirations, sometimes people tell you what they think they want to be.
Right, we all want to weigh 150 pounds.
Hay: Yes, that's right, certainly my case, my aspiration. Sometimes it's just frankly faulty memory. We ask people questions, such as have you bought Greek yogurt in the last month, and then we'll put that alongside the actual behaviors and purchases in the store and find that sometimes, it can be 90% inaccurate.
The reality is we don't go around with perfect information in our head about our lives. So we have customers where we track that purchase behavior and we ask them, we get their permission to ask them about actual behaviors and aspirational data. And when you put the two together or the three together, it can be immensely powerful.
So to what extent can you actually match actual people to the numbers and the things they say? To what extent is that useful, and how do you help your customers navigate these potentially treacherous privacy waters?
Hay: Two things. One is just because you can do with data, doesn't mean you should do with data. We are very clear on some no-go areas around what you'd analyzed, what you'd look at. When we do a research panel, for example, we ask customers to elect themselves in and we actually pay them in the form of points and rewards for sharing that data. Too much of the data in the digital ecosystem is being used to do things to customers, when we believe that data should be used with and for customers. If you can't look the customer in the eye and explain what data you've used, why you've used it, and why it's a benefit to them, then you probably shouldn't be doing it.
Doing surreptitious things with data, we wouldn't consider that to be a great strategy for growing your business. We tend to operate in the spaces where we know there's so many opportunities that you can clearly collect data from customers, manage that, and use it to create a better customer experience and do so in a way that customers understand and recognize they get value. The great thing about big data is that it's coming more quickly to more places, and therefore, creates more opportunity.
Now the irony of this is actually, as data opportunities grow, the more organizations are struggling to know what to do with it, what matters, what doesn't. Some of the cultural aspects of organizations—how to use systems, how to use data, how to measure the right things, how to reward the right things—become even more necessary, and actually, harder as data grows. So our opportunity is to help organizations use data and put in place the relevant systems and culture to get the best from data and technology.
Certainly the real-time piece is a huge game changer, particularly in the digital world. But ultimately, when it's either served up to the customer or served up to a decision-maker in a business, it's got to be simple, logical, and actionable. Then for the customer, it's got to make sense—it's made my life a little easier, a little better, a little cheaper. But the timeliness of it and the size behind it has just sort of come on thousands of fold in the years that we've been working.
From where you sit, is digital disruption the sort of looming menace to existing companies that it’s made out to be? Are we talking about a sea change, or just a small tidal shift?
Hay: I think, honestly, it's a bit of both. You look at the way that music's bought or the way that people are creating business models and services, tracking your pizza from the moment it's made to the moment it's delivered. I think it's an opportunity. Stores are still massively important in the retail world as a place to see, as a place to touch, as a place to experience. You haven’t heard Apple complaining about its store network and what a terrible millstone it is around their neck.
So actually, there's increased demands on us to be flexible, to be quick, to be agile, smart, all of those things, which you always had to do to win someone’s business—you just need them a bit faster and better than you did in the past. To do that, you need a bit more science and data and creativity. You need left and right brain, whereas in the past, maybe there was a different balance.
It feels like it's a paradox in that, because sometimes you need mega-speed, as in a real-time world, and sometimes, actually, you need to work your way through some of the complexity. You need to be more thoughtful. And so with all these things, it places demands on leadership. That's always been the thing that differentiates great businesses from the average ones.
Where do you think the balance lies in terms of the balance or conflict between intuition and data? And if there's a conflict, how do you, as CEO of a company, resolve that yourself?
Hay: It's a great question. When we talked about the source of the data and the volumes of data, there used to be one truth in data. Now you can create arguments that tell you it's night through data, and someone else can tell you it's day. That's where expertise comes in, because people use data and statistics selectively to build an argument. So you’ve got to know about sample sizes, data sources, claimed behavior versus actual behavior, and some of the modeling techniques used. I suppose one of the ironies of my business is I probably use intuition as much today if not perhaps even more so than I've done in previous years, as the complexity and volumes and speed of data increases. Sometimes, I've got data telling me it's night, and I've got data telling me it's day, and I have to make a choice. In the end, we still rely on human beings to do things with the data. And obviously, sometimes machines are doing it, but even then, the machine is still dancing to the tune you set it to, to the strategy you set, the parameters, the economics, the budgets, the strategies.
In the course of your engagements, have you had customers find unexpected realizations about new potential revenue streams? You know, has a retailer discovered that maybe they should become a telephone company?
Hay: Well, actually, Tesco's history is a great example of where data has taken them into new markets, because in essence, what took them into telecoms, what took them into banking and other markets, is really understanding what their customers spend on food in their stores.
Sometimes you do use data to turn heads and upend clients' understanding of their strategy. People may think that their business should be very [customer-] acquisition focused, but if you can prove to them the value of their existing customers, the value that's walking out the door by not retaining them, and the growth and headroom that still lives within even their best customers, then obviously you can change their strategy and their focus.
