Special Report: Shopper Labs

The concept of product manufacturers using shopper labs is not new, but we're not talking white-walled conference rooms and a few nondescript shelves here. CPGs today have access to a spectrum of increasingly sophisticated research tools, including customizable mock stores and 3-D virtual labs."A shopper lab can be anything that attempts to create a representation of a live shelf," says Dan Boehm, North American head of retail and shopper for TNS, a Kantar Group unit that owns one of the leading virtual reality providers, Red Dot Square. "There's value in all the different approaches. It's a matter of knowing strategically what you're trying to accomplish and creating something as real as possible."Various leading CPG companies – including Coca-Cola Co., Kimberly-Clark and Procter & Gamble – have put considerable resources into internal shopper lab research facilities in recent years.Coca-Cola no longer delegates its lab research to outside providers, thanks to the opening of a Shopper Experience Innovation Center (SEIC) near its Atlanta headquarters in May 2012. The beverage giant has brought those capabilities in-house to the 15,000-square-foot SEIC, which Shopper Marketing visited last fall.Coca-Cola says that, since the SEIC's opening, it has not only produced better insights and brought packaging and display concepts more quickly to market – a process that takes anywhere from three to 18 months, depending on the complexity of the project – but has also vastly improved its cross-functional approach to developing innovative merchandising solutions. "Being in the same room with our retail customers and seeing how people [i.e., shoppers] interact with our products, packaging and merchandising has been a tremendous bridge for collaboration," says Ron Hughes, director, shopper experience innovation. In the past, a merchandising concept may have been exposed only to specific retail customer teams before it was sent into the marketplace. Now, Hughes says, "route drivers, channel people, brand people and merchandisers are all coming through our doors and providing feedback on solutions. There's more knowledge in the funnel."For retailers, the SEIC offers an unfiltered look at shopper behavior – thanks to extensive video monitoring capabilities, an observation room and two large prototype testing areas that can be outfitted with realistic shelving, lighting, signage and features as diverse as a grocery store deli counter or a full-scale QSR menu board."We've had customers tell us that some of the things they observed [at the SEIC], had we told them, they would not have believed," says Hughes, explaining that most retailers are interested in the "in-store connections" that SEIC shoppers make and how they interact with point-of-purchase materials. "They want to make sure we are converting customers and not just adding clutter to their store."Real and Virtual: The K-C WayAt Kimberly-Clark's 6,000-square-foot Innovation Design Studio, tucked away from the main building at its Neenah, Wis., campus, a virtual reality test center sits alongside a physical lab environment. The latter accommodates up to four double-sided gondolas comprising 16-foot sections of a generic store.K-C was one of the pioneers of virtual reality testing back in 2004 and has since done enough comparative testing to gain a nuanced understanding of both methods. "If we want to understand path to purchase or consumer engagement for various merchandising scenarios – did shoppers notice things on shelf? – virtual reality is a great fit because we can create different scenarios and adjust the parameters with a few clicks," says Kim Greenwood, senior manager of virtual reality.In general, online virtual testing affords greater flexibility, can draw on much larger sample sizes with wider demographic representation, and can be done in far less time than a physical store test. It also allows for early-stage testing of packaging or merchandising materials before anything is manufactured or produced. The costs are generally considerably lower for virtual reality in single category tests but can become prohibitive to account for all the interactive elements needed for total-store testing.On the other hand, physical store environments play a critical role in testing concepts that involve a sensory component of sound, touch or smell, says Bala Mallela, K-C's senior director of category and shopper insights. Additionally, small elements such as a price tag change may not be detected on virtual reality screens, he says.Variables that typically get tested with virtual reality include changing up placement of shopper marketing materials, different merchandising configurations and adding new product mixes to the shelf. "With [virtual reality] we're able to take actual store sales data and restocking data and simulate different states of shelves at different times for a retailer, understand switching and walk-away behaviors," says Greenwood. "Let's say it's Saturday morning and a mom is doing a stock-up trip at a large retailer for essentials like paper towels and toilet paper. What does she do when she can't find the products she's looking for on the shelf?"In 2011, K-C conducted extensive virtual reality testing for merchandising concepts following the launch of its U by Kotex brand – and enjoyed a 60-basis-point increase over sales expectations as a result of recommendations implemented from those tests. Specifically, K-C learned that vertical brand blocking would be an effective way to leverage the product's deep black packaging and generate shelf presence. Retailers continue to employ that tactic today.P&G Shapes the Tide Pods LaunchResearch conducted over a period of 18 to 24 months at Procter & Gamble's Beckett Ridge Innovation Center in Cincinnati similarly uncovered key insights that impacted strategic decisions on the launch of Tide Pods laundry packets.P&G began with several rounds of virtual reality testing to determine which package design iteration for Pods would be most noticeable on shelves, and which would best connote a laundry item. Once the "fish bowl" design was in place, both virtual and physical store testing were used to identify the key elements of brand equity and how those would impact merchandising and point-of-sale materials."There are significant shopper barriers you have to overcome when you are launching a new form, and that required us to use many research tools," says Marshall Haine, leader of retail innovation for global operations at P&G. "The first moment of truth has been key to the success of Pods. Some of our retail customers have been particularly good with their merchandising in communicating the benefits of the brand."Predicting Shopper BehaviorThe conventional wisdom that shopper labs are a necessary tool to gauge purchase intent is based on a simple premise: What people in surveys or focus groups say they will buy often isn't borne out by their actual purchases. Of course, no shopper lab can fully replicate a real-world shopping trip (or match the scale of an entire store). For one thing, the psychology behind an actual trip mission is different than that of a study participant. Plus, the controlled environment of a shopper lab cannot account for the effects of crowds, noise and other exogenous store factors.That said, shopper lab researchers tout some pretty impressive case study results. Among them is In Vivo BVA, which claims to be the world's largest independent purveyor of shopper labs with 19 testing facilities around the globe, including U.S. sites in New York, Los Angeles and Chicago. In Vivo BVA USA came on line in early 2010 and has since worked with major CPG companies including Kraft Foods, PepsiCo, General Mills, Sara Lee, Nestlé and Campbell Soup Co. The research projects may be completed in as little as five weeks or in several months and cost anywhere from $20,000 to $200,000, according to John Traynor, president and chief executive officer of In Vivo BVA USA.Traynor says that the company's predictive power is greatest in two areas: package tests and new product introductions. "When we've done in-market validation of package tests against Nielsen or IRI data, using our STM [simulated test market] model to adjust for things like merchandising or advertising, we've come within 3% of determining the effect of the package change 90% of the time," he explains. "For new product introductions, we can predict first-year sales volume with the same level of accuracy at a plus or minus 9. We don't have enough data yet for merchandising, but we think we are in the same [predictability] range as new products."The Case for VirtualVirtual reality providers say they can achieve similar results. "We've done many side-by-side studies with in-store behavior and consistently seen high correlations," says Rich Scamehorn, chief research officer at Chicago-based InContext Solutions. He co-founded the 3-D online virtual reality company in 2009 after several years of running both virtual and controlled store testing at General Mills.In 2012, InContext Solutions tested the impact of a layout redesign in the candy aisle for a major grocery and mass retailer. When it later compared the results with actual sales data provided by the retailer, it found an average correlation of 0.9 or higher with dollar or unit shares of products, brands and sub-categories with a total of 40 products.Those kinds of success rates, along with advances in its software technology, have led to the firm's growth and recent expansion into categories such as consumer electronics, telecommunications, and home and garden. "We're developing more complex simulations that incorporate total store interactivity, simulated shoppers and scripted avatars," says Scamehorn. "It's all to create a more realistic shopping environment."Despite the growing adoption of virtual reality testing, some experts remain skeptical over the validity of its quantitative findings (see sidebar at right). Scamehorn acknowledges that some of his firm's own clients are wary that respondents in a simulation would behave as they would in a store, though he says the concern comes more from newer users of virtual reality who've yet to be exposed to its predictive value."The main risk is that respondents might find the environments too interesting or fun at first, leading them to 'play' before beginning the task they've been asked to do," he says. "This results in longer time and more interactivity in the environment, which can skew the results." To counter this, all participants must complete a training tutorial on how to navigate the environment and are questioned extensively about their shopping habits. "We're very careful about screening our observations for any unnatural behavior."