By Brian Kilcourse, Managing Partner
12/4/2007
At the recent Customer Data Security Executive Meeting co-sponsored by NRF and RSR, Catalina Marketing EVP and CIO Eric Williams gave a keynote entitled “Beyond PCI: The Challenges and Opportunities in Using Customer-Specific Data to Create Value for Consumers.” Eric discussed how his company is working with retailers and CPG manufacturers to understand consumer choices, and to use that understanding to fine-tune incentive offerings to be more relevant to consumers, while at the same time protecting each consumer’s privacy. As an anecdote during his presentation, Eric mentioned an example of how consumer market basket data can be used to create value. Catalina worked with one retailer to try to identify an early warning indicator that customers were becoming less loyal. After several scans of their huge database, Eric and team thought they had come up with the one item that is the best early indicator that a customer was starting to wander. That item? Milk.
Catalina has come a long way since its inception in 1983. The company now supports targeted marketing to the point-of-sale for over 22,000 stores in the U.S. alone. According to Eric, the company has the purchasing histories of approximately 135 million consumers, and this information is used to help client companies (retailers and manufacturers) offer targeted and relevant incentives to their customers, to build the loyalty bond so important in retail.
Catalina has been using data mining capabilities for five years to help retailers and manufacturers reach the “right” consumers early for new products, in the knowledge that if the product or the retailer selling that product can garner an early market share lead, that advantage will last a long time – often at least 18 months. In a followup conversation last week, Eric said, “It wasn’t until about two years ago that we brought in a team of dedicated technology people for data mining. What we found out very early is that (obviously) consumers are creatures of habit. We also figured that there was logic in the idea that if you can be the first person to get a consumer to try a product, that they’ll stay loyal to that item until they switch for some reason outside of the your control. “
Catalina implemented data mining technologies to see if they could in fact predict consumers’ purchases in a category before they purchased anything in that category. From that effort, Catalina got a request from a retailer who wanted to know if the same technology could be used on the other side of the loyalty cycle, to predict people that were about to stop shopping in a store. According to Eric, “We said, ‘what the heck’. We took a look at a bunch of consumers that had stopped shopping in their stores – we could see that either as declining number of trips or declining dollars per trip – and we threw about a year’s worth of data into the data mining tool and said ‘tell us something interesting’. We asked the system to do a correlation – pre data to post data. What came out of that was a result indicating that you could in fact predict a high propensity that the consumer would stop shopping that store based on a single item – fluid milk.” The retailer subsequently ran a special promotion directed to those customers who specifically weren’t buying milk anymore.
Eric hopes that this story will replace the now-hoary urban legend about baby diapers and beer, and frankly, we hope so too.
Behavioral marketing has always been constrained by technology’s ability to handle the volume of the data and the reiterative nature of the analysis, and Catalina has frequently “pushed the envelope” in this regard. The company has implemented SAS Enterprise Miner over a Netezza “data warehouse appliance” environment. The data warehouse is over 100 terabytes large, making it one of the largest databases on the planet. Says Eric, “We’ve done a tremendous amount with the data mining solutions and with general logical extensions of what we’ve learned to develop a trigger set for offers. This is a hot ticket for our clients since the real scenario that if you can be the first to get a consumer to try your product, that becomes the standard that others are compared to.” Now Catalina runs on average between 8 and 10 scorings each day, for retailers and manufacturers. The technology is so fast that Catalina can score the entire datastore “in minutes,” according to the CIO.
Eric notes, “The technology of data mining is becoming prevalent in the marketplace, at companies like Catalina and many of the banking companies are using these solutions to help them to perform analytics and therefore predictive analysis of consumer behavior.”
Indeed. The opportunity to use customer-specific data to create compelling value for consumers is very real for Catalina and its clients. Back at the Customer Data Security event, Eric spoke to the audience about how retailing is really headed back to the value proposition of 100 years ago, when retailers knew their customers by name, delivered products when they were needed, and focused on “fresh.” The difference now is that digital information is empowering retailers and their manufacturing partners to make relevant value offerings on a huge scale.
|
