Watching as you shop
Retailing: Big shops are using elaborate technology to monitor and influence the behaviour of their customers
“Shoppers are not just being monitored by cameras and sensors—sometimes speech is recorded, too.”
IN A recent British television commercial, John McEnroe, a veteran tennis star, is seen racing his long-time rival Bjorn Borg around a Tesco supermarket. After arriving in the checkout queue just ahead of his opponent, Mr. McEnroe is distraught to see an employee appear and open another till, giving Mr. Borg the upper hand. The advertisment ends with a reference to Tesco's “one in front” initiative, which aims to reduce waiting times by opening additional tills whenever there is more than one customer waiting at a checkout. This scheme has been running for many years, but Tesco recently updated it by investing in customer-monitoring technology supplied by Irisys, a British company.
The system, called Smartlane, has two elements: sensors by the doors count the number of people entering and leaving the shop, and sensors by the tills work out how fast the queues are moving and how many “shopping units”—groups of people, such as families or couples, who will make one transaction between them—are standing in each queue. (Shopping units are identified using an infra-red sensor that tracks the trajectory and behaviour of “hot blobs” when they enter its field of vision; this approach is 95% accurate, says Chris Precious of Irisys.) All this information can then be used to predict how many tills will be needed up to an hour in advance and monitor average waiting times and queue lengths. In Tesco's case, a line longer than one shopping unit triggers the opening of another till. Sir Terry Leahy, Tesco's boss, said last year that the monitoring system had reduced waiting time for customers and helped to boost the firm's profits.
Tesco is not the only retailer using technology to keep tabs on people in its stores. Mr Precious says Price Chopper, an American grocery chain, is in the process of introducing Smartlane; and Marks & Spencer, a British shopping chain, uses Irisys's infra-red sensors, as does Abercrombie & Fitch, an American fashion chain. Brickstream, a supplier of customer-monitoring systems based on dual-lens cameras, which are more accurate than single-lens cameras in tracking and counting applications, counts big retail chains including Toys “R” Us, Office Depot and Walgreens among its clients.
Rather than monitoring queue lengths, however, Brickstream's system, called BehaviorIQ, is used by retailers to gather data on where their customers go, where and how long they stop, and how they react to different products. This information can then be compared with the store's transaction log to determine the effectiveness of store layout, product fixtures and other variables. In-store designs and marketing campaigns that work can be identified and improved upon, and those that do not can be replaced with something more profitable. Retailers, it seems, are buying the idea that by watching customers while they shop—what might be called “retail surveillance”—and redesigning their shops accordingly, they can get people to spend more.
On the shop floor
One of the largest customer-monitoring projects undertaken so far is PRISM (or “Pioneering Research for an In-Store Metric”), a collaboration between Nielsen Media, a market-research firm, the In-store Marketing Institute, based in Chicago, and a consortium of consumer-goods manufacturers and retailers including Coca-Cola, Procter & Gamble and Wal-Mart. The project, which completed its first big trial in September, involved the use of sensors at the entrance and exits and in some of the aisles of 160 stores across America. These sensors recorded data on customer-traffic patterns, to which was added further information recorded by human observers. By comparing the resulting data with sales information, it was then possible to gain insight into shoppers' behaviour.
Preliminary results from the PRISM programme included the finding that two-thirds of shoppers who made a visit to the salty-snacks aisle made a purchase, but the people visiting the dairy aisle were far less likely to buy something. When observational data on the number of children present during a shopping trip was taken into account, the results were even more detailed: the PRISM results showed that although children took part in only 13% of food-shopping trips, shoppers bought more when children were with them. Purchase of seasonal items was two and a half times more likely with children in tow, but surprisingly there was no correlation found between the presence of children on a shopping trip and higher spending on sweets.
With this kind of fine-grained data, says George Wishart, the head of Nielsen's in-store division, retailers can adjust store layouts and staffing levels accordingly. Although retailers think they know how to position products in a way that will maximise sales, says Peter Hoyt of the In-store Marketing Institute, “the science of adjacency is not a science.” With PRISM, he says, the gut feeling that peanut butter should be placed on a shelf next to jam in order to increase the likelihood of a combination purchase can be tested directly against sales data.
Best Buy, a big American consumer-electronics chain, takes this a step further. It uses Brickstream's BehaviorIQ cameras to collect data on its customers' behaviour patterns and has used this information to divide them into five customer types—young technology enthusiasts, suburban mothers, affluent early adopters, family men and small-business owners—who want different things. Each group is then provided with an area of the shop designed to meet their needs.
As the technology improves, it is getting easier for other retailers to do the same sort of thing. VideoMining, for example, a spin-out from Pennsylvania State University, uses a combination of video cameras and software to capture and analyse thousands of hours of footage of customer behaviour. As well as providing information about customer behaviour, the firm's software can also determine each shopper's sex, age range and ethnic group with an accuracy of around 80%, says Jeff Hershey of VideoMining. The result is an “activity map” which can help retailers to identify the factors that drive higher sales. The system can even be programmed to detect the characteristic behaviour of shoplifters.
Inevitably, all this raises concerns about privacy. Counting customers with motion sensors as they enter a supermarket is one thing, but sorting them into social or racial groups using video analysis is another. Mr Hershey responds that none of the footage or information about individual shoppers is stored. Besides, says John Greening of the Medill School at Northwestern University, customers are generally willing to give up some personal information if they feel they are getting some benefit in return. Rather than seeing customer monitoring as an invasion of privacy, he considers it a means of making shopping a more pleasant experience for “time-compressed” (ie, busy) consumers. These systems, he says, are “not a matter of Big Brother”, but rather a means of gaining insight and “understanding motives”.
Shoppers are not just being monitored by cameras and heat sensors, however; sometimes their speech is being recorded, too. Recordant, an in-store monitoring company based in Atlanta, provides shop assistants with digital-audio recorders, to be worn around their necks, which record all their conversations with shoppers. At the end of each day the recording devices are plugged into special docking stations and the recordings are uploaded to Recordant's headquarters for analysis. Its software trawls the recorded dialogue for particular words or phrases without the need for any human input in order to provide accurate information on employee performance and customer behaviour.
Recordant says that trials in 40 locations have proved the value of the technology: one retailer was able to show that there was a 300% increase in sales of products recommended by shop assistants to customers. By providing precise measurements of this kind, the technology allows retailers to refine their training programmes and evaluate the effectiveness of promotions. They can even collect “competitive intelligence” by programming the system to search for the names of rival retailers, says Chris Etter, the chief executive of Recordant.
Unlike systems based on video cameras or sensors, Recordant's technology is required to notify customers explicitly that their conversations are being recorded. To this end, the recording devices include a small video screen which displays a message informing shoppers what is going on. Mr Etter admits that both employees and customers have expressed worries about privacy and have shown some resistance to the devices. But he likens this to the initial disquiet over the recording of customer-service telephone calls, something that is now widely accepted. Besides, he says, the system can provide protection for employees who get involved in disputes with fractious customers.
As useful as customer-tracking systems may be for exonerating harassed staff or minimising shoplifting, however, their primary objectives are clear. In an age when traditional “bricks and mortar” retailers face stiff competition from online merchants, any means they can use to get customers into their shops and buying as much as possible is seen as a worthwhile investment. Long before the advent of e-commerce, Wal-Mart's founder Sam Walton set out his vision for a successful retail operation: “We let folks know we're interested in them and that they're vital to us—’cause they are,” he said. Customer-monitoring systems are underlining just how interested, and just how vital.