Online CRM Key for E-Commerce 2001SEATTLE -- The flavor of e-commerce this year is all about making money and an improved online customer experience is critical to that end, according to Ken Burke, president/CEO of Multimedia Live, who addressed a workshop here yesterday at the Direct Marketing Association's net.marketing show in Seattle.
"Keep in mind that customers may not tell you what your flaws are," Burke said, adding that complaints, when they do come in, are very loud.
Burke touted intelligent selling and personalization as key ingredients to creating and operating a Web business by optimizing all selling opportunities.
"It's really talking about selling the right products to the right person in the right place at the right time with the right message at the right price," he said.
This shouldn't be difficult for catalogers, who have an advantage over their bricks-and-mortar counterparts. For instance, catalogers can identify their customers. Plus their customers are used to buying direct.
"Catalogers should get 18 percent of their sales online," Burke told an audience of 40 direct marketing executives at the morning session. "For retailers, it's a little bit more of a problem. They normally get between 0.5 percent to 10 percent -- at the high end -- of their sales online."
A key way to improve online customer experience is through predictive modeling. Such a system uses a statistical calculation, or a probability theory, to determine the likelihood of a user taking a certain action, like abandoning a shopping cart.
"Shipping is the No. 1 reason" for shopping cart abandonment, Burke said, adding "people use carts for storage and the right calculation [of products in the basket]."
Another reason for cart abandonment is credit card rejection. Burke said that, for some reason, MasterCard users showed a higher shopping cart abandonment rate than rival Visa.
But while predictive modeling is good for highly evolved sites, basic personalization -- registration and account information, and name recognition -- is good as for entry-level personalization.
Collaborative filtering, on the other hand, is a flop, according to Burke. This method of personalization aims to automatically determine the target consumer group from click stream and other data and then applies that to situations. This includes recommendations like, "Customers like you also like…"
"It's a failure in our industry," he said, pointing out that a consumer that may have bought a gift for a baby may continuously get served promotions in that area. The site does not recognize that the consumer has bought a gift and not a product for self-consumption.
Burke reserved his highest praise for rules-based personalization. Under this method, the online retailer builds target consumer groups and then applies business rules based on a user and a situation. The format calls for If/Then relationships.
"I think this is the most powerful personalization [method]," he said. "It creates target consumer groups that allows for execution of messages."
But for any effective personalization, direct marketers should start profiling using a cohesive data collection strategy. Data should be gathered from many different sources and aggregated into a universal customer database.
To enable this, three pieces of data should be married: Online data, which includes click stream, online purchase history, tool use, survey information, wizard data, referral information and online campaign use; backend data, such as customer data, offline and online purchase history and external data sources; and external data like buying patterns, demographic information and data from Experian and Abacus.
Burke recommended best practices for an effective personalization strategy.
· The online retailer should specify what events will affect a customer profile.
· The strategy should define user groups based on customer information or site navigation behavior.
· It should define business rules for merchandising and messaging to each user group.
· It should test the effectiveness of marketing, merchandising and messaging campaigns by producing pertinent metrics and indicators.
Things that can be personalized include site messaging, images and displays, product sort order, product selection, category structure, pricing, campaigns and promotions, tools usage and e-mails.
For instance, a customer group that has been to the site but never purchased should view a custom site. Using present and past click stream data, the site should show particular items and deliver specific content to encourage sell-throughs.
For a customer that visits the site and buys sometimes, Burke suggested a custom site, with targeted offerings, cross-sells and content. Using a combination of previous purchase, present and past click stream and offline data, the marketer can then offer bundled goods, provide messaging and create promotions to move targeted products.
And, in the case of a customer that has been to the site and buys often, the ideal site experience is a custom site, special promotions and targeted offerings, cross-sells and content.
Again, using previous purchase history, present and past click stream and offline data, the merchandising should reflect special incentives, additional content and the request for more data via a survey.