TDWI Upside - Where Data Means Business

EBay's Revamped Customer Database Creates Insight

EBay's Customer DNA database is a lot like the customer journey on steroids. It gives the online giant a unified view of all its users -- even those who buy and sell from different accounts.

EBay's Sridhar Paidi was on hand at TDWI's recent Executive Summit in San Diego to talk about Customer DNA, the online giant's new customer data repository.

In Paidi's telling, eBay's Customer DNA sounds a lot like the customer journey on steroids. Its 65,000 data points give eBay a unified view of all its users -- even those who buy and sell from different accounts.

It also helps address challenges that aren't unique to eBay, including correlating registered users with "Guest" checkouts, identifying fraud, and promoting cross-sell or up-sell opportunities. Even though Customer DNA, like any ambitious analytics effort, is still a work in progress, it's a quantum improvement over what came before it, Paidi told attendees.

Defining Users to Optimize Customer Interaction

"We didn't know much about our customers, we just didn't have much detail about them. That's why we started building [Customer DNA]," he said. "Lucky for us, we had a good, influential vice president. He came up with this new vision and he said, 'Let's do something about this. Let's use data and analytics to optimize every interaction with every customer.'"

Thus was born Customer DNA. Paidi and his team started with an effort to link eBay user IDs across all accounts and services. This was a harder problem than it might seem.

"This is one of the biggest problems. The definition of 'user' changes with the use case. For example, I have a couple of accounts on eBay. I have one for selling and one for buying. How do we link those accounts?" One strategy is to link on the basis of a common email address, which is how Paidi and his team started. Many users who maintain different account names for their buying and selling activities nonetheless use a single email address.

There were other problems, however. What about a household that -- as with Amazon Prime -- shares a single account. Who is the eBay "user" in this case? How can eBay identify the individual members of the household and market to them? Should eBay market to all of the members in the household? What about children or teens? EBay is no longer just an auction website, Paidi noted; a growing number of shoppers use it as a retail site via its "Buy It Now" option.

Perspectives on Customer Experience

Because eBay defined data points for every conceivable customer interaction, it was able to collect a massive amount of information about its users, from their clickstream history to their checkout behavior. With enough data (and sufficiently differentiated buying patterns), it becomes possible to create a reasonably accurate portrait of the composition of a single user's household.

Paidi and eBay took this even further, however. The portrait they'd created was from eBay's own perspective; what does this particular customer's experience look like from her own perspective?

"Checkout is telling you a story from Checkout's perspective. When you flip the story around and look at it from the customer's perspective, it's very different," he said. "Instead of looking at all [of] the checkout [data], look at how many check outs ... in certain categories. I may be checking out in just certain categories," Paidi continued, explaining that even though a customer might browse in many different categories, he or she may tend to buy (to check out) items in only certain categories.

"When you have these data sets it becomes quite easy to link them together and answer questions such as who are the users who had X number of checkouts with N number of visits ... or who are the users who have a lot of customer service [interaction]," he said.

Measuring Engagement

EBay's Customer DNA database also permits it to measure customer engagement. The online giant is analyzing basically everything its users do, from their first to their most recent transaction, the average period between transactions, and (in the case of newer users) even before this: their first, second, and most recent visits to eBay's site.

Factor in customer service contacts and other data points, Paidi says, and you have an excellent profile of how engaged your users are.

"This gives you a picture of a user that ... [also] becomes a powerful data set that all the groups [in eBay] can use. We started with users that we didn't know much about ... now we know their age group, what their demographics are, what their interests are, what their behaviors are, and even more," he said, explaining that this makes segmentation analysis much more insightful.

"Now we're able to use these data sets to let data scientists create sophisticated models of data." This is possible, Paidi argued, only because eBay was able to create a highly detailed view of its customers. "These analytics go into many things, [and] are used in many ways," he concluded.

"All of our marketing efforts ... are customized using [these analytics]."

TDWI Membership

Accelerate Your Projects,
and Your Career

TDWI Members have access to exclusive research reports, publications, communities and training.

Individual, Student, & Team memberships available.