Before the phrase “big data” became commonplace, dunnhumby was helping retailers and brands understand shoppers and grow sales by extracting value from massive data volumes. Founded in 1989, dunnhumby is on a new mission to democratize data and make sophisticated analytics available to all. In April 2017, the London-based company lured Guillaume Bacuvier away from an 11-year career at Google to lead a new phase of growth as CEO. He spoke with Retail Leader about personalization, sku rationalization and the data science company’s evolving business model.
Retail Leader: Most everyone has heard of dunnhumby, but in simple terms how do you describe what dunnhumby does?
Guillaume Bacuvier: We help companies and primarily retailers extract value from their customer data by applying science to it. Companies produce a lot of data in their day-to-day business, especially in retail, and we help them use that data to better understand their customers, make better decisions and better serve their customers. We don’t necessarily touch all of the data sets a retailer would have, we are really focusing on data that has some type of customer connection.
RL: A retailer would use dunnhumby in addition to providers of other types of shopper insights?
GB: Potentially yes. A typical interaction we have with a retailer very often begins with helping them understand and audit all their data sets. There is usually an assessment of the IT and data infrastructure and some recommendations about how to best organize and set up the data so that it can be fully exploited. From there we help retailers use data to answer fundamental questions about what their customers are doing and why they are doing it and creating reporting and business intelligence that is customer centric. For example, answering with great insight simple questions about why foot traffic or sales are dropping in a store or region. Or helping them build customer segmentation that is smart and always live and relevant and can be updated almost every day. We also help them optimize pricing and assortments and personalize communications and promotions using a lot of advanced science. What dunnhumby is all about is activating retailers’ data to extract value from it in different ways.
RL: Thanks for clarifying. There is a lot more data available today than when dunnhumby was founded and it was common for retailers to complain about not knowing how to put data from point of sale systems to good use. Now we have this proliferation of data from structured and unstructured sources making for an even bigger challenge.
GB: The volume of data produced in a retail business has increased by multiple orders of magnitude. The other challenge that many retailers face is they were early in using IT and building systems that have aged so there is a big disparity in their technical infrastructure and the quality of their data that puts them at a disadvantage relative to new upstarts natively built as IT companies with cleaner data sets.
The other thing that has changed is twenty five years ago you could find companies who had a lot of data and could accept that they didn’t quite know what to do with it all. Today there is a recognition that data usage is a key competitive advantage. One of the key points I made at our North American Partner Summit is that the best retailers and businesses need to stop looking at the data in their systems as a byproduct of their core business and more as a critical asset in its own right that they monetize and treat strategically.
RL: You also talked in your presentation about redefining customer-first retail in a data driven age and mentioned “the new retail.” What does that mean?
GB: That is a term we stole from Alibaba founder Jack Ma. He talks about the new retail as the integration of online and offline channels, supply chain and data as one system. He puts data as one of four things that are key to a new retailer and views the data he is collecting and processing as important as stores and web sites. We agree with that. Increasingly what will make a retailer successful is not just having stores in the right places, or the right products or employees providing the right service, it is having the best data set from which they can optimize their operations across the board. Data is one of the key pillars of any retailer’s strategy and will determine who is successful in the long run.
RL: I don’t think many people would dispute that or dispute that data is what drives being customer centric by executing personalization.
GB: The case we are making based on the work we have done with hundreds of retailers is that everyone in the organization from the board room down to the store front uses data to understand the customer and manages the business and make decisions on the back of data that tells them more about customers. Starting with the board room, usually the metrics and reporting and therefore the tools by which any company makes a decision are typically the result of the data they can get their hands on and the systems and organizational design they have. Therefore, it is very rare for the key metrics an executive team tracks to be centered on customers. Keeping an eye on performance by customer segments, how your most loyal customers are doing and slicing the data by customers doesn’t come naturally to businesses because that isn’t the typically business are typically organized.
It is important to make sure that you use data to get that customer centric mindset and the data that backs it is available at every level of the organization from the boardroom to the store front. A lot of what dunnhumby does is enabling that, providing the entire organization with the insights, the reports and the tools to make that a reality.
RL: Oftentimes it’s the merchants or marketers who are the biggest users of data so it is interesting to hear you talk about board level usage of customer data.
GB: With dunnhumby’s largest, most successful clients we typically have board level engagements. Some of our clients have redefined the metrics and KPIs by which the executive team manages the business to put more customer data into their hands it because it rarely comes naturally to any business. We are focused on how to make customer centric data available to all and therefore every decision that is taken, whether by a store clerk or the CEO, is driven by customer data. Through data, you try to create the notion that you personalize each interaction to each customer at scale.
RL: That sounds very challenging, especially for a large retailer operating in multiple geographies.
GB: It is easier said than done especially in a multichannel environment with physical stores. We try to replicate as much as possible what is capable online in an offline environment.
RL: Personalization can mean different things depending on the category or type of store. If a shopper is in a larger store they expect a broader assortment.
GB: True, though there is a counter argument that is reflected in the work we have done for quite a few retailers. In very large store formats there is a trend toward assortment rationalization because having too many Skus is counterproductive especially if some of those Skus are perfect substitutes for one another because you are creating more confusion for customers. The challenge is finding the optimal tradeoff between offering something for every client that walks into the store without offering too much because it becomes counterproductive.
RL: Isn’t finding the perfect assortment a perpetual challenge for the industry?
GB: With the data sets that retailers have today, whether it is rich loyalty card data or tokenized basket data, you can draw insights about what are the products you cannot afford to remove versus those that are such perfect substitutes for one another they are arguably duplicates and therefore how do you create the optimal assortment.
RL: Some stores offer very limited assortment in one category but you can go down another aisle and there is a more extensive offering.
GB: Some of that reflects the tension in the business around whether retailers are putting the products in the store that customers want or is the product assortment the output of trade negotiations with suppliers. We have told the story of Tesco’s turnaround, they are our oldest customer and shareholder, over the past three years. A lot of what has gone on involved them moving away from a merchandising approach to squeeze the maximum juice out of suppliers at the expense of the products and prices customers really wanted and it has allowed them to regain market share and grow.
RL: I’m glad you mentioned Tesco. They own a percentage of dunnhumby but that doesn’t preclude you from working with other retailers.
GB: We have two separate relationships. One is a shareholder relationship and the other is fairly standard commercial relationship like we have with other clients. There is no exclusivity so we could in theory work with Tesco’s most direct competitors in the U.K. They have been very supportive of us because in order for dunnhumby to be successful we need to operate as a normal business.
RL: Is it a case where the larger the universe of companies you work with the better it is for Tesco because the deeper the overall market insights?
GB: That’s right. Obviously Tesco cannot get its hands on any data specific to our other clients, but the aggregate expertise of our team is much stronger from working with many retailers. Likewise, as anyone who is working on software algorithms processing data knows, the more data these algorithms are trained against the better they become so it is in the interest of everyone that dunnhumby has as many clients as possible because that makes our science better over time.
RL: What does the next phase of growth look like for dunnhummby?
GB: Historically, if someone wanted to work with us and take advantage of our data and expertise it was expensive and it was complex. Realistically it was only something a larger retailer could afford. Our strategic direction at a high level is to enable companies of more varied shapes and sizes to access our science so we have created a data science on demand model that can be accessed by a much more varied set of businesses. Our go to market model has already started to shift from a model that was high touch, with a lot of consultants and multi-year engagements, which we will still employ with some retailers, to one that allows companies to access our data science and expertise through a lighter touch model. The long term vision and thesis is that even if you are small independent store, using data to better understand data your customers and optimize pricing and assortments is important too. We want to enable any retailer to access the same level of sophisticated data science as what we are able to provide to Tesco, Whole Foods or others we work with today.
RL: You spent 11 years at Google, a very well regarded company, why make the switch to dunnhumby?
GB: There are similarities in philosophy and culture because Google also believes you can extract a lot of value from data. What attracted me to dunnhumby was I saw the potential to make this a really fascinating company if we are able to achieve the science as a service vision. The potential with recent technology evolutions is to create a platform of data science that we could address a much larger market and. It seemed like a transformation story for the business that was very appealing. We can really make the company an amazing story in the coming years if we play it right.
RL: Can you extend into other retail channels beyond food and consumables?
GB: It even goes beyond retail. Part of the strategic planning we have done revolves around the question of can we broaden our addressable market and where can we go beyond our sweet spot, which is grocery centric retail. What we do also applies to other retail segments. Sometimes the offering and the science has to be adapted.
RL: Any segments you can single out?
GB: E-commerce is an area where we have done work, but haven’t had a structured offering. People assume because they are data and tech natives they are good at it and have nailed every problem but our experience is that is not true. We have a number of pure play retailers in our portfolio today but we think there can be more. E-commerce is definitely an area where we want to invest.
RL: Data is data. A lot of the same techniques are used to extract value from it regardless of channel.
GB: Any business that collects a lot of data about its customers is interesting to us. That is our filter when we look at sectors we think we can play in.