Marketers are spending big on increasingly sophisticated e-CRM technology as they dig down deep into customer behavior in the quest to make the online shopping experience more relevant for each shopper-and to see bigger sales result. But as the experience of some marketers shows, the best-designed e-CRM initiative can be effectively torpedoed by a basic element that may be overlooked: the quality of the customer data underlying that effort.
“Companies can spend millions on CRM initiatives or customer analytics and customer intelligence exercises, but what they really need to do first is focus on whether their data is strong,” says Atique Shah, vice president of CRM/marketing technologies at GSI Commerce Solutions Inc.
Shah should know. With a client roster of 54 that spans The Sports Authority and Chanel, and the charge of online marketing on each client’s behalf, e-commerce service and technology provider GSI Commerce has had lots of experience in collecting customer data online: registration data, preference information, transaction information, survey information, clickstream data. The customer data gathered for each client and housed separately in a massive digital warehouse was a potential CRM goldmine-if segments of the data could be defined and extracted in a way targeted enough to guide marketing and merchandising decisions.
“Back in January 2001, we realized we needed to work with CRM more effectively to deliver value for our client base. We were looking for a solution that could help us realize better the value in that data,” says Shah.
For help in accomplishing that, GSI two years ago turned to technology provider SPSS Inc. and its Clementine data mining tool. That software application runs on top of CRM software from provider E.piphany Inc., which encompasses call center applications and its own clickstream analysis tool. The integration of the data mining tool with the CRM suite and its legacy system was a largely painless six-week endeavor, but before it could start spinning the refined customer data into gold, the implementation revealed that GSI had a problem to fix first: data quality.
“One of the biggest problems a database marketer can face is the quality of data. If the data is not segmented and manipulated properly, I don’t care what the architecture is-it’s not going to fly,” says Shah.
The implementation revealed unsuspected “disaster zones,” in the data, Shah says, for instance, multiple listings for the same customer of the same store. That led GSI to launch what Shah dubbed “Operation Data Validation,” in which inaccurate, incomplete or duplicate customer data was routed out, with some help from SPSS’s data mining tool as well as tools attached to its data warehouse. Data cleansed, the e-commerce provider started to see that, with strong data as the foundation, e-CRM does indeed pay off, as aggregate results from its several sporting goods clients shows.
GSI Commerce’s marketing team groups customers in four segments according to purchase history: bronze for customers who make a one-time purchase, silver for two purchases, gold for three and platinum for four or more purchases. Some of its online campaigns to those customers weren’t segmented, but when it realized in looking at reports from the database that the lifetime value differed among groups, it started targeting content and offers accordingly.
“You don’t treat the platinum customer the same as the bronze customer,” Shah says. “Some of the testing we did on our sporting goods sites was phenomenal in the sense that the same e-mail campaign that used to deliver a good 1% click-through, an average industry rate, with a 0.5% conversion, is now delivering over 10% click-through with a 6% conversion rate after we reworked the message and redelivered the content, working with the creative department. Across sporting goods it really got a huge lift from 1% to close to 7% conversion.”
The key to producing that kind of gain was not just the CRM applications from E.piphany and the data mining and segmentation technology from SPSS, but also the effort that went into ensuring that customer data-the building blocks of any e-CRM program-was cleaned and validated before being used to build the program.
Return to e-sender
The implementation of e-CRM software from Blue Martini Software Inc. three years ago similarly revealed unexpected problems regarding customer data-specifically, faulty e-mail addresses-at digital storage product manufacturer and direct marketer Iomega Corp. Initially, Iomega acquired the software to give itself greater flexibility in presenting price lists to larger-volume, end-user purchases.
But it also allowed marketing to avoid having to ask IT, whose staff had shrunk during the recession, to program routine marketing campaigns. “That gave us a system that allows merchandising staff to do their work without having to write any code or get involved with IT,” says Monique Fraser, senior manager of information technology at the company.
That kind of flexibility, which made the marketing staff more efficient, was also a factor in Iomega’s decision to use Blue Martini’s tools in bringing management of its e-mail campaigns in-house. Previously, Iomega had assigned e-mail marketing to an outsourced provider. “Using Blue Martini meant that people who want to send e-mail campaigns have the capability of going into the system, selecting from a segment of our user group who is going to receive the e-mail, and designing and dispatching it themselves without involving IT,” says Fraser.
The Blue Martini software “opened our eyes” to the quality of Iomega’s e-mail distribution, Fraser says, revealing that the company had been paying for distribution of a fair number of e-mails that never arrived because they bounced back. “Previously, when we’d ask the service provider to send out a certain number of e-mails-say, 2 million-we’d look in a soft way to see what the impact of that campaign was against what we were paying them to do that work for us,” says Fraser.
When Iomega brought the management of its e-mail campaigns in-house, the e-mail list itself remained housed in the provider’s data warehouse. When the company started requesting customer segments to launch e-mails intetrnally, it now was able to see that the lists were full of malformed e-mail addresses which had gone unidentified in the database for some time.