Harnessing Big Data

Meet "fully profiled" shopper Mary. Within a month's time, she buys a pair of running shorts online that reviewers say are non-chafing, and also 3 pounds of pasta at her local grocery store. She searches recipe sites and Pinterest for healthy pasta dishes.These behaviors, along with Mary's full scope of online and mobile activity – checking weather forecasts, using a fitness app, reading articles on carbo-loading – all point to her intention to run a marathon, which is soon confirmed when she signs up online.While Mary is not real, the possibility of knowing this much about an individual shopper is very real. And bringing these data points together could enable shopper marketers to make relevant offers and talk to Mary differently than they do Sally, another heavy pasta user who buys it because it's cheap and filling. Unlike Mary, Sally has four kids and not a lot of money. Marketers might also know of her intention to take the kids to Disneyland, lose weight, become pregnant again or plant a vegetable garden.Using transaction and shopper data, in-store video monitoring, geo-tracking, real-time social media feeds, Internet browsing history, local conditions such as unseasonable weather or a crime wave, and a whole host of other structured and unstructured data – along with computer programs and artificial intelligence to make sense of it all – brands and retailers could reasonably predict what individual consumers and their families will do, need and buy.But marketers have yet to figure out how to harness this "Big Data." Like stars in a galaxy, the data points already exist, but marketers need to "connect the dots" so that "constellations" – patterns, pictures and actionable insights – start to emerge, says Laura McCorvie, senior vice president, customer growth, Kraft Foods Group."In many ways, we've only just begun to realize the potential of Big Data as a means to effectively engage shoppers today," says Deanie Elsner, executive vice president, chief marketing officer, at Kraft.Kraft has built "deep, sustained relationships" with consumers through its magazine, website, Kraftrecipes.com and iFood Assistant app. "But while we've been interacting with consumers for some time," Elsner says, "we're now becoming more adept at connecting what we're seeing in social media, our own consumer and shopper feedback, our shopper marketing program results and primary research. We also have the chance to integrate what we have with retailer-owned data."We've just scratched the surface of connecting these rich data sources to drive promotions based on individual shopper preferences and behavior."Making Sense of Big DataBig Data is commonly characterized by the three V's – variety, velocity and volume. The data comes from many sources and in various structures and formats; it comes at us quickly; and we're talking terabytes, petabytes or even greater amounts of data. Every day now, humans create 2.5 quintillion bytes of data – so much that 90% of the data in the world today has been created in the last two years alone, according to IBM."Successful Workplace" blogger Chris Taylor, in his December post "The Non-Hype Definition of Big Data," favors the "more pragmatic" Wikipedia definition: "a collection of data sets so large and complex that it becomes difficult to process using on-hand database management tools." By defining Big Data by the challenge of using on-hand tools, Taylor writes, "we acknowledge that the term means something different to everyone."Brand marketers and retailers looking for shopper insights see Big Data as a potential gold mine, and with good reason. A 2011 McKinsey study suggests that by using Big Data to the fullest, a retailer could increase its operating margins by more than 60%. Retail giants like Kroger are investing accordingly, partnering with Big Data crunchers like dunnhumby to exploit the vast reservoir of POS and loyalty card data already at their disposal.Still, it seems we're a long way from a 60% gain. In order to see such a payoff, marketers would need to use Big Data to its fullest. But what would that entail? And what needs to happen to bring our Mary scenario to life?POS Data and BeyondPoint-of-sale data is understandably the starting point. "Retailers certainly have a lot of valuable, transaction-level data," says Chris Ciccarello, senior director of customer analytics at ConAgra Foods.From a CPG perspective, "certainly anything that represents real, non-aggregated consumer behavior is valuable – definitely frequent shopper data or just transaction-level sales data where there is no shopper card," Ciccarello says. "There's a lot there because it's real, very granular shopper behavior. And because it's shopper and transaction level, you can do things like segment and target shoppers likely to be most responsive, understand market baskets (what consumers buy at the same time to inform what items to include in a shopper marketing event), track shoppers over time for long-term impact of your programs, and gain insight into how your shopper marketing is working" in terms of bringing in new buyers or making existing buyers more loyal.Paul Kramer, chief executive officer of Westport, Conn.-based shopper agency CatapultRPM, agrees that POS data is "good" for targeted marketing within a retailer's own "ecosystem," but combining it with syndicated data offered for sale by the likes of precision marketing agency Catalina is "better." Together, they would reveal insights about Sally as an individual, including her price sensitivity and how often she stocks up. But there's no qualitative data (lifestyle or psychographic information), "which you need to truly address the needs of the family," Kramer says. You can target heavy pasta users, for example, but you can't address Sally's or Mary's particular needs.What's best, Kramer says, is to "take lifestyle data and overlay it with the two other types, so you can fine-tune your messaging to better engage" individual households.Enter Data CompaniesTo that end, some major data providers are playing in the space. Dallas-based consumer data company Epsilon's recent acquisition of CatapultRPM could mean that Big Data is poised to transform shopper marketing in a big way, Kramer says. Epsilon, which also hosts the electronic platform for Walgreens' Balance Rewards loyalty program, gathers lifestyle data via consumer questionnaires addressing everything from pets and music preferences to specific health conditions.Fort Worth, Texas-based customer analytics company Buxton works with CPGs to identify who their best consumers are. Drawing from 250 databases of demographic, psychographic and behavioral data, Buxton helps identify potential new customers "who look just like them" in a particular trade area, says Bill Stinneford, senior vice president of account management.Jim Harold, general manager of the Consumer Industry Group at the enterprise data firm Acxiom Corp., Little Rock, Ark., says Acxiom's overall data strategy is to "move consumers from unknown to fully identifiable with their network of friends from social media," along a "customer image spectrum" progressing from anonymous to "fully profiled," like Mary. Online consumers are assigned a unique ID for recognition and traceability, but their identifications and data are anonymized for privacy. "Where we're going is the linkage between online and offline personas in an anonymous way," Harold says.Glittering like so many unreachable stars, tons of lifestyle data sits on various devices and across many platforms, from Mary's fitness app that links to her Facebook page to her positive review of her non-chafing shorts. Organizations are using the open-source software program Hadoop to "dump massive amounts of unstructured data" to make sense of later, but so far only a fraction has borne fruit, says Harold.Reluctance to Share"Harvard Business Review" blogger Alex "Sandy" Pentland, referring to Big Data in general, wrote that the biggest barrier to Big Data applications is "getting the data out of silos and into a form" where it can be used. "This isn't just a garden variety, get-your-departments-sharing kind of corporate problem," wrote Pentland, director of Massachusetts Institute of Technology's Human Dynamics Laboratory and the MIT Media Lab Entrepreneurship Program. "It's more difficult than that, because with Big Data, it is typical that no one company owns all the data you need" and collaboration is required.In the shopper marketing world, "POS data is guarded very closely by retailers," CatapultRPM's Kramer says, and manufacturers must "buy into their programs utilizing that data."Right now, "Retailers have the preponderance of data. And they don't want to share it until they understand it," says Bill Pearce, a marketing professor at the Haas School of Business at the University of California who previously held C-level and senior marketing positions with Del Monte Foods, Campbell Soup Co. and Procter & Gamble.However, retailers currently do not have the capacity to make sense of their data or enrich it with lifestyle and unstructured data, Pearce says. So, for mining actionable insights, "third-party players are going to be important." (Pearce serves on the board of one such vendor, RichRelevance.)Rightfully, retailers view their data as valuable "and don't want to give it up to manufacturers," Kramer says. But sharing data and insights "provides for more relevant, targeted messaging" to consumers.According to Kraft's Elsner, there's also "rich data" at the manufacturer level that is "generally not intertwined with retail insights at the individual shopper level. Manufacturers, retailers and shoppers will all benefit if we can find ways to rethink what has traditionally been a siloed approach to data management."Third-party vendors and data management platforms provide a "safe haven" or "neutral environment for data storage, analysis and optimization where retailers aren't giving it away and manufacturers aren't giving theirs away," Kramer says. Having established clear objectives and performance accountability for any co-branded programs, both parties will benefit from the richer database, as well as Big Data-driven analysis of the impact of shopper marketing programs, events and offers."From the retailer's perspective, a CPG has so many dollars to spend to drive sales of their product at [the retailer's] location, and those dollars drive traffic to the store where [shoppers] will buy other things," Buxton's Stinneford says. "The retailer wants to secure as many of those dollars as possible, but it's competitive and highly scrutinized. So there has to be a tempting proposition to take to the CPG."Working with Buxton, retailers can tell CPGs who's buying their products at a particular location and what those buyers have in common, as well as point them to others who fit the profile, Stinneford says. "Then, we can do a PIN-coded offer that needs to be scanned when it's redeemed for a response analysis to identify the subset who actually responded to that offer."Challenges AheadWith the hype focused on Big Data's potential for targeting "fully profiled" shoppers like Mary, less is said among shopper marketers about "the opportunity Big Data provides us with to test things – campaigns, displays at retail – virtually in real time, change variables, get feedback, adapt accordingly and test again," says Craig Elston, senior vice president of insights and strategy, The Integer Group, Denver. "We can live in a much more iterative world versus a linear development world."Shopper marketers can test Big Data-driven hypotheses on a small scale and gauge the success before a major rollout, says Scott Setrakian, managing director, Applied Predictive Technologies, Arlington, Va.Big Data can also be used to optimize supply chain management. "There are data sets out there that could help us do a better job forecasting, because like many companies, we've missed revenue opportunities by having out-of-stocks or we've incurred high costs by transporting product" where demand exceeds supply, McCorvie said at a recent Path to Purchase Institute event. "If we can better predict demand in real time, we can have product at the right place, at the right time."At a Supply Chain Conference in February, Ahold USA announced it would share data from its stores (including Stop & Shop, Giant-Landover, Giant-Carlisle and Martin's) with CPGs through a third party in order to improve product availability and consumer satisfaction.Challenges presented by Big Data include "the puzzle of correlation versus causality," Setrakian says. The more massive your volume of data, the more "statistically significant" correlations it will yield, though most aren't meaningful connections. For example, a Big Data set shows you a correlation "between the amount of broccoli someone buys and their hat size," he says, "but the correlation is not causal."There are also issues of privacy and shopper-centrism. "One of the risks associated with Big Data is its focus on what marketers could do and can do. What often gets lost is what we should do," says Matt Nitzberg, dunnhumby's executive vice president of communications and media. "You need to have a shopper-centric orientation and see to it that the shopper is excited and delighted to hear from you. That's really where the art meets the science."Target famously raised some eyebrows – and ire – for employing an algorithm to tell whether a woman is expecting a baby, based on her past purchases. A targeted mail offer to a pregnant teen "outed" her condition to her father.For ConAgra's part, "I see more and more opportunities to get big data sets in-house, and I see it getting easier, cheaper and faster to manipulate," Ciccarello says. "It can only enhance our ability to understand how shopper marketing works and how to make it better. We have an internal 'Shopper Marketing Measure, Learn, Change process,' and we can leverage data to make it better."Going forward, some believe all but a few third-party vendors will disappear, leaving only those who mean business to produce results. "The best parallel I can give you is the Internet boom, where 90% of the companies that sprang up around the hype went away," Harold says, "and the Amazons and eBays emerged and changed the way America shops."

P2PI MEMBERS

JOIN P2PI

You May Also Like

The Future of Ad Targeting – Why Retail Media is the Solution, Not the Problem

AD Retail Media’s Transition In-House