Thumbnail image for data.jpg

Perhaps the only universal truth about big data analytics is that it is a newfield and everyone is still learning. That was true Monday as data scientists,quants and business intelligence managers gathered in San Francisco toswap ideas about what's working.

While retail corporations dominated Monday's schedule, the PredictiveAnalytics World conference also drew financial executives, marketing managersand others who know all too well that the success of their companies may dependon making better use of the river of data flooding in from social media,transaction records, web surfing, retail outlets and other sources.

Two early presentations came from executives at Sears Holdings and eBay, twoof the companies that have led the industry in growing successful onlineoperations. Their presentations were minutes apart, and both stressed theimportance of experimentation, but they covered widely divergent areas fromdifferent perspectives.

The Sears Approach

In a presentation for Sears Holdings, the 10th largest retailer in the US,Robin Glinton,  director of Sears' data sciences lab in Silicon Valley,said the company is trying hard to ensure all customers from all its brands canbuy any item either online or in-store and pay for it in any way they'd like.

"Instead of focusing on bringing  in new customers, we focus ongetting our existing customers to buy more from our other brands," he said.His presentation showed how Sears creates 3-D models based on customer segmentsand testing them against various spending "threshholds" and valuepropositions to optimize offers that yield the maximum benefit for its brands.

For example, a deep discount may attract many shoppers. But what if you wantto entice a particular type of shopper to buy more. What is the minimum theymust spend, and what must the value be to them? Through careful  analysisand testings, Sears has found it can generate a 20 percent increase in emailopen rates with a 15 percent increase in offer redemptions.

Learning Opportunities

Over at eBay

"If you go on eBay  right now, you'll see hundreds, if notthousands, of experiments going on," said Gayatri Patel, director of eBay'sdata strategies. It's all part of eBay's effort to make more relevant dataavailable to employees at all levels. Using a variety of BI tools likeMicrostrategy, Tableau, SAS and R, the company provides dashboards forexecutives and portals for others managers who want to create "personalworkspaces" to analyze any of the data in eBay's system.

EBay's goal over the next two to three years, she said, is to increase howquickly the data can be analyzed and used to know more about the customer inreal time, near time and post time.

For example, she gave the example of a customer who loses in an eBay auction.The company would like to let that customer know  about similar auctionsthat are still going on,  but first it must understand if the customer onlywants new items, wants a particular brand or only wants used goods innear-perfect condition. That's a big challenge for when selling everything fromantique furniture to fashion-forward handbags.

Traditional retailers may be worried about online stores like Amazon, butmany are working with eBay to test different geographical markets or customersegments. EBay shares its data with its retail clients to help them evaluatewhat works best, she said.

Came to Learn

"I want to hear how cutting edge analysis is working,  hear whatnew vendors have to offer and to network with my peers," said ShyamGottapati, director of analytics at Visa. While the lessons from Sears or otherretailers don't translate directly into  lessons for financial servicesproviders, he noted the retailers are in the forefront of establishing bestpractices in big data.