Big Data — the data generated by millions of online, mobile and other interactions every day — helped Barack Obama understand the electorate and regain the presidency last year. Can it help food retailers, too?
That certainly seems to be the sentiment among retail analysts and vendors such as IBM, SAP, Oracle, SAS, Manthan Systems, Teradata, Symphony EYC and others who sang Big Data’s praises at the National Retail Federation’s Conference & Expo in New York last month.
Retailers, too, are veering in the direction of using more and more data in making decisions about promotions, assortments and supply chain performance. Notably, Canadian retailer Metro Inc. , Montreal, which operates 430 corporate stores, has partnered with London-based Dunnhumby to leverage data from loyalty programs as well as digital data inputs to create a new “customer-centric” strategy.
Jeff Roster, a retail technology analyst with Gartner, Stamford, Conn., reported that in a Gartner survey of retailers, 29% said they have already invested in “technology specifically designed to address Big Data challenges” including servers, software, analytics and consulting, while 12% plan to do so in the next year and 29% in two years.
Read more: BevMo’s IT Is ‘Up in the Cloud’ 
For Richard Mader, former executive director of the Association for Retail Technology Standards, and now president of Mader International Consulting, Bernville, Pa., Big Data is something retailers are going to have to embrace. “You better understand Big Data and get involved with social media, or somebody is going to come and steal your customer,” he said.
Consumers’ activity on websites and social media, as well as on their mobile phones and on video sensors in stores, among other sources, represent the new digital data inputs for retailers, along with traditional sources such as transactional and loyalty data and syndicated data.
The volume and velocity of Big Data are impressive. Wal-Mart Stores  by itself collects more than 2.5 petabytes (2.5 million gigabytes) of data every hour from its customer transactions, according to Andrew McAfee and Erik Brynjolfsson, researchers at MIT’s Center for Digital Business, who wrote about Big Data in the October 2012 issue of the Harvard Business Review. Each petabyte is the equivalent of about 20 million filing cabinets’ worth of text.
Unsurprisingly, there are a host of technical challenges associated with Big Data. For one thing, the digital component is often “unstructured” as compared to the structured databases used by retailers. In addition, traditional business intelligence engines alone may be insufficient to tap into Big Data, and advanced analytics are needed to leverage the data for applications like forecasting, optimization and personalization. It may also be necessary for retailers to hire the new breed of “data scientists” to get the most out of Big Data.
Read more: Woods Cuts POS Losses With Video Analytics 
On the other hand, the declining cost of computing is making data-intensive technology more affordable, noted McAfee and Brynjolfsson. With that, a multitude of new technologies supporting Big Data are coming on the scene. Sears, for example, has set up a “Hadoop cluster” a group of servers controlled by an open-source software framework called Hadoop.
Cloud-based systems may also provide retailers with a way of handling Big Data, noted Mader.
Metro has partly addressed these IT and personnel challenges by partnering with world-class consumer data experts Dunnhumby. “The challenge of Big Data is that we need analytic solutions and a team to do that,” noted Marc Giroux, chief marketing and communications officer, Metro, who spoke at an NRF Conference session. “That’s the strength of our partnership with Dunnhumby — we can leverage Dunnhumby’s global presence and practices and that allows us to increase our speed to market.”
Big Data also raises concerns about consumer privacy that retailers will need to address. (See “Data Deluge Creates Privacy Issues ,” SN, Oct. 22, 2012).
Change in strategy
Metro’s partnership with Dunnhumby over the past few years is a good example of a retailer attempting to leverage a wide range of customer data to drive a personalized relationship with its best shoppers as well as a customer-centric approach throughout the organization.
The change in strategy at Metro was prompted by the surge in retail square footage in Canada, notably the growing clout of Wal-Mart and the entry of Target , said Giroux, “So the fight is on to keep our loyals [loyal customers].”
Dunnhumby, with its deep experience in loyalty marketing gained through similar partnerships with Tesco  in the U.K. and Kroger in the U.S., is well positioned to help Metro make the leap into Big Data. Throughout all of its 17 markets worldwide, Dunnhumby looks at data from 400 million customers, representing sales of $600 billion, said Marc Fischli, chief operating officer, international markets, Dunnhumby Canada, Montreal, who also presented at the NRF session.
Read more: Retail Group Formed to Address NFC Technology 
The customer data used by Dunnhumby and Metro come from several sources. It starts with the Metro & Moi (me) loyalty card program established by Metro in Quebec in 2010, as well as well its longtime Air Miles program in Ontario. Both reflect Metro’s revamped brand strategy, known as Mon Epicier, or my grocer, emphasizing proximity and personalization.
With the Metro & Moi program, shoppers get a point for every dollar spent (more for promotions), and earn $4 for every 500 points garnered. Those earnings add up to a reward check mailed every three months along with personalized offers. The reward mailer has a 90% open rate and generates a palpable “buzz” in the store, said Fischli.
But the online world and the development of digital technologies “has also given us lots of different data sets we can start to interact with traditional data” and “do this faster and even better,” said Fischli.
Dunnhumby is being “quite selective” about its digital data choices, said Fischli. “It is an interesting challenge for a company like ours to see which data in the big amount of new data available we should use.”
Read more: ARTS to Launch Work Team for Charity Standard 
So far, Dunnhumby has tapped Facebook content and Internet data such as click-throughs. On Facebook, Metro is willing to have an open conversation with its shoppers about sensitive issues like customer service, said Giroux.
Customers whose Facebook issues are addressed may become “advocates” for the Metro brand, which opens up “new opportunities on the data analytics front,” said Fischli. “What is a customer like that worth? You can start personalizing your interactions with those individuals even more so to make best use of their willingness to actually advocate our brand.”
Dunnhumby also leverages its BzzAgent subsidiary, a social marketing company that puts products in the hands of consumers who share their opinions via Facebook posts, photos and videos, as well as blogs and other avenues. “You would have some very interesting data pieces coming through their agents and you can use that to enhance the overall picture,” Fischli said.
All of Metro’s collected data is employed to “personalize every interaction we can with customers,” such as the targeted promotions sent with reward checks, said Fischli. Metro is also using the data to “make better decisions across the business,” added Giroux.
The personalization of fliers is another key component of Metro’s strategy and Metro has also focused on making store assortment more tailored to the needs of its most loyal customers.
The combination of loyalty rewards and personalized offers has led to “a good increase in the frequency of visits and the average size of the baskets by our most loyal customers,” said Giroux. In fact, since embarking on its customer-centric strategy, Metro’s stock price has “outperformed the sector and competitors,” noted Fischli.
On the other hand, data programs like Metro’s are not without a significant investment. Fischli advised investing between 1% and 2% of top-line sales “to have a successful loyalty program.”
Sidebar: Big Data in Japan
SUNNYVALE, Calif. — Japanese retailer Trial, which operates 150 Wal-Mart-like stores, has developed a fast data engine to manipulate Big Data, and is now in the U.S. marketing that system and others through its subsidiary Trial Retail Engineering (TRE) here.
The data engine, called Smart, is “like a relational database, and it’s fast and cheap, using batch processing,” said Yosuke Noda, chief operating office for TRE. It crunches numbers “at the UPC level,” he added.
Read more: Tech Effort Transforms A&P: CIO 
The system supports a variety of applications, including basket and clustering analysis, a geographic information system (GIS) and customer targeting.
The data that Trial is capturing in Japan includes POS, loyalty and demographic information. “We are combining those to decide what should be put in [newspaper] fliers in targeted regions,” said Noda. Trial is also determining the right product assortment to attract shoppers in particular regions as well as the impact of competition and promotions on sales.
TRE is marketing its database as a cloud-based service, noted Noda.
|Suggested Categories||More from Supermarketnews|