The Wal-Mart trial involves IBM, Sprint and EZface. It is being tested at 10 very different locations—with four units at each store—based primarily on demographics and somewhat less on connection speed. A list of the 10 locations includes each store's distance from signal source, with all stores in strong 3G signal areas. An early list of potential stores included a 4G candidate, but it didn't make the final cut because the chain decided to stick with the more common 3G for more reproducible results.
The tests are being done in three areas identified as rural general market (Maryville, MO; Emporia, KS; and Decatur, TX), three pegged as suburban general market (Fishkill, NY; Hamilton, OH; and Sioux Falls, SD), two considered higher income (Amherst, NH, and Pembroke Pines, FL) and two labeled empty-nesters (Payson, AZ, and Livingston, TX).
(See related Wal-Mart stories this issue: Wal-Mart: "It’s Time For Chip-And-PIN In The U.S." and Should Wal-Mart Digital Signage Use Near-Time News, Weather, POS Data?)
The trial offers a wide range of barcoded cosmetic containers coupled with a digital camera and a screen. The customer looks into the camera and takes a picture of herself and then scans the barcodes to virtually apply various makeup products to her face. The customer is asked to remove glasses, brush hair away from her face and remove any hats before taking the photo.
After the customer's selections are made and applied, the image is saved and E-mailed to her.
During the trial, the images are not linked to the customer, who is not identified, nor are customer E-mail addresses saved. That means there is no testing of the CRM implications, such as whether the machine causes an action (either that day or on a subsequent visit). Given that Wal-Mart doesn't use loyalty data anyway (it seems to thoroughly detest even saying the words "two-tiered pricing"), it's hardly a surprise.
There's also no attempt—again, limited to this 90-day trial—to allow purchases at the point of the demo. That certainly may change if Wal-Mart ultimately decides to deploy. For the trial, the only return-on-investment information is on a store-wide basis. If cosmetics sales increase sharply at the 10 tested stores, that's the best ROI Wal-Mart will have.
Psychologically, the chain is hoping these devices will push sales for reasons beyond efficiency. "Today, people may try one or two things (makeup) and they tend not to look at the full face," said one official involved in the trial. "With this system, the potential exists for much stronger increased sales."
The trial is using Sprint CDMA for data connectivity, which made it easier for Wal-Mart to do the tests. Sprint CDMA didn't require any of the test systems to interact with the chain's LAN, thereby bypassing questions such as "Do I want to put in a separate WiFi network in the stores?" said that official, adding "With this, all we have to worry about is power."That same official said the decision to put four units in each tested store was based on the nature of the cosmetics area: two units for each area and another two on the opposite side of the aisle. "We may have Maybelline on the right and another brand on the left," he said.
Another early consideration was putting a high-resolution printer, which would likely have better color-matching capabilities, at each counter. But concerns about keeping the printers supplied with paper and ink/toner made it seem too hassle-oriented. "If I want to make it as hands-free as possible," he said, "sending a digital version sidesteps the burden on the store."
The idea of using technology to better sell cosmetics is hardly new. Macy's is now toying with self-service centers (with admittedly little tech) and, at the other extreme, Japanese mega-chain Mitsukoshi has been experimenting with RFID tags to accomplish something similar to what Wal-Mart is trying to do with barcodes. If it works perfectly, RFID would theoretically be easier for consumers because it sidesteps barcode scanning. Customers in an RFID environment could simply grab devices representing blush, eyeliner or lipstick and start designing away, while a barcode trial requires scanning with each new element before it can be virtually applied.
All of these screen-based efforts still have to deal with the not-so-makeup-friendly lighting in a typical department store, in addition to getting a display to both show accurate colors and interact realistically with store lighting. Even Hewlett-Packard has been trying to figure out a way to deliver accurate colors to retailers, using less-sophisticated mobile cameras.
The Wal-Mart trial, while advanced, still has to deal with the day-to-day challenges of a convenient cosmetics test. For example, it needs to ask consumers to remove eyeglasses. Whether that's for glare or, more likely, to expose the full face for virtual painting, it will deliver a look that may not reflect how the customer will look in the real world. Then again, glasses may make some profitable eye treatments less necessary by obscuring the enhancements, so removing them may not be a bad thing margin- and revenue-wise.
Part of Wal-Mart's objective is to see how merged-channel (the final stage after cross-channel, which itself is one step beyond multi-channel) a cosmetics trial could be. Thanks to the E-mailed images, nothing about the trial is limited to in-store, Web or mobile, theoretically allowing the trial to test all three channels.
"It's intended to be a multi-channel experience, whether in-store, at-home or using a mobile device," said Craig Velliquette, a client executive for IBM's Wal-Mart team. The consumer "is able to retrieve the image no matter what device she is using," he said.
The trial is not limited to traditional makeup; it also includes hair-coloring treatments and other facial helps. "It's not just about virtual makeup," Velliquette said. "It's providing customers shopping assistance for a variety of looks, including day and night looks."