Using Data to Stage-Manage Paths to the Prescription Counter

Every day, five million people walk into a CVS.

What they do next â€" pick up a gallon of milk, linger in the hair care aisle, make a beeline for the pharmacy â€" is where things can start to get interesting.

Like many retailers, CVS is collecting and analyzing as much data about its customers’ habits as it can. The entire industry is mining information to understand what people want and how to get them to want more.

Wal-Mart, the nation’s biggest retailer, is analyzing social media chatter to come up with new products. AutoZone is using software to change its inventory in a particular store based on the kinds of cars people in the area drive.

And CVS, the nation’s second-largest drugstore chain, is crunching internal data and industry figures to come up with new aproaches to such things as what products it should stock where, and what special offers customers should get.

“The data isn’t completely new, but it’s taken a long time for retailers to be able to do much with it,” said Matt Waller, chairman of the supply chain management department at the University of Arkansas and chief data scientist at Orchestro, a data analysis firm based in McLean, Va. “Processing power and storage space has become so cheap that a lot more can be done, and of course there’s better algorithms.”

At CVS, the first goal was to find the best customers. The chain had always assumed that it should tailor its stores to frequent customers who were buying shampoo, Advil and other front-of-store, or nonprescription, items.

With data analysis, though, CVS rethought that approach.

It turned out that the best customers were people who visited the pharmacy frequently. These are largely people with chronic conditions, said Helena B. Foulkes, chief health c! are strategy and marketing officer for CVS Caremark. “There’s a high frequency of visit, which is often driven by the fact that they’re taking multiple medications â€" diabetes, heart disease,” she said.

CVS executives named their hypothetical supercustomer Beth, a 50-year-old woman who handles prescriptions for her children, her husband and her parents.

The chain next analyzed what else shoppers like Beth bought when they came in to refill prescriptions. It turned out shoppers used the stores differently depending on whether they were in a city or in the suburbs.

In urban stores, customers treated CVS like a general store. They bought drugstore staples, but purchase data indicated they also wanted grocery items, prepared snack foods, household items like toilet paper or batteries, and baby products. So in cities, those categories were expanded and a new “On the Go” food section with items like sandwiches and precut fruit was added. Products in the seasonal aisle were cut (goodye, shelves of beach towels).

In suburban stores, where a CVS may share strip-mall space with a Walmart, a Safeway and a Home Depot, a different pattern emerged. People were not picking up light bulbs and milk much from CVS stores. Instead, they were buying health and beauty products alongside their prescriptions. CVS is now redesigning suburban stores to give more prominence to those areas.

“Ultimately the question is, How do you design your store for those who are most valuable to you?” Ms. Foulkes said.

Data is also informing how the stores are arranged. Oral care was one puzzling area. CVS could never figure out where, exactly, to put toothpaste and floss (with grooming products, first aid or cosmetics?). Purchase data indicated that customers were most often buying those items alongside beauty products, so CVS reshuffled the stores to put dental products close to the beauty aisles.

CVS also applied data to another problem: how to get customers to adhere to their presc! riptions.! Pharmacy makes up about 70 percent of store sales, meaning that the chain has to get people to take their medications if it wants to maximize revenue.

Once CVS merged with Caremark, a prescription-benefits company, in 2007, it could tackle this problem a little more broadly. Caremark tracks whether a customer is filling a prescription anywhere, not just at CVS stores. (To keep a separation between CVS and Caremark, CVS pharmacists can’t see the names of other pharmacies or the cost of prescriptions there, CVS says.)

CVS found that one-third of consumers stopped taking prescribed medicine after a month, and half stopped after a year â€" and that is for chronic diseases, where medications are meant to be taken for at least two years, on average, Ms. Foulkes said.

So CVS began an automated program of texts, e-mails and phone calls to remind people to refill prescriptions. It also added an alert to customer profiles, so pharmacists could stop customers when they were in the store picking upanother prescription and ask if they would like to chat about the one they weren’t taking as prescribed.

CVS created talking points for its pharmacists based on the therapeutic class of the drug. For instance, people tend to go off cholesterol-lowering medications because they do not feel any different. Now, pharmacists counsel customers that though they may not feel different, they should continue taking the drug. With antidepressants, pharmacists warn that sleep and stomach issues are common two to four weeks in, and the patient should keep on going.

The other big use of data is for personalized offers. CVS has a 15-year-old loyalty program called ExtraCare, and about 70 million people have used the card in the last six months. Personalized offers have been a focus for the chain: last holiday season it sent 117 million personalized offers, mostly through printouts on the bottom of receipts, up 72 percent over the previous year.

With all those years of purchase histories, CVS has! unearthe! d certain patterns. Some have to do with what’s in the customer’s shopping basket: someone who buys skin care and hair care products at CVS is most likely buying cosmetics somewhere, even if it’s not CVS. Others have to do with timing: people usually buy toothpaste every five weeks.

Now, CVS can send offers or coupons to try to change buying habits. For instance, it might send a discount on mascara to the person buying lotions and shampoos at CVS, or a particularly aggressive discount on toothpaste if five weeks has elapsed without a repeat purchase. Or, if someone tends to spend $15 each time he visits, CVS will send him offers meant to increase his purchase total to $20 the next time.

The company also found that brand-specific offers, like an offer to save $2 on a L’Oréal cosmetics purchase, were not redeemed as frequently as general offers. Merchants now rarely press vendors for specific offers, instead creating category discounts, like a $2 savings on any cosmetics purchase.

Ad as merchants look at what they carry, the data helps them figure out what to discontinue and what to keep. Something might be a slow seller, for instance, but if the customers Ms. Foulkes describes as “high value” buy it every now and then, merchants will usually keep it in stock.