SOME DATA TRAILS FROM HOLIDAY SEASON LEAD BACK TO RETAILERS
by Ed Colet
I wrote a column that appeared two weeks ago on the electronic data trails left by online shoppers this holiday season. Well, it seems that a good portion of those trails lead back to the retailers' sites in the form of returns and exchanges, and less than satisfied customers. Two lessons can be learned from following these trails. One is that we can learn, through data mining, about the reasons customers are dissatisfied with online purchasing. And second, it becomes clear that an integrated system is important for success whether it be an e-commerce site, or in the larger picture, an implementation of a data mining solution.
Depending on when and how growth is measured, the statistics on this season's holiday revenue from online shopping are generally quite impressive. PC Week's January 4 issue reports statistics collected by the Boston Consulting Group that show that compared to 1997, holiday sales were up 230%, had an annual sales growth of 200%, with an average online order up 6% to $55, and projected annual sales for 1998 to be $13 billion.
Despite these impressive revenue figures, not all shoppers were satisfied. A survey commissioned by Jupiter/NFO shows that only 74% of shoppers were satisfied, down 14% from 88% of satisfied shoppers six months earlier. (But another study by VISA, reported by Reuters, January 20, 1999 claims that 98% of shoppers are satisfied).
The conclusions from both surveys are that web-sites need to improve customer service issues in order to have satisfied customers. A lot of effort goes into attracting customers to a site, especially in light of the fact that it's been reported that approximately only 5% of visitors to a site actually make a purchase. Other than attracting users to a site, an equally important objective is to have these people consistently return as loyal customers. Without attention to customer service a lot of the effort to attracting users to a site may be lost. (The column two weeks ago alluded to one-to-one marketing philosophy that characterizes personalization in which the intent is to build and develop customer loyalty).
Careful analysis, through data mining and/or other means, can shed more light on what affects customer satisfaction. According to the Jupiter/NFO study, the reasons for dissatisfied customers can be attributed to merchandise, shipping and handling, and speed (at the site). Data mining can potentially discover patterns that may be useful. For example, do customers that are not satisfied with merchandise spend more or less time at the site? i.e. Does it take awhile to finally find a product? Or are products bought quickly and ultimately not what they promise to be?
Dissatisfaction associated with shipping and handling issues is easier to speculate about. There is the added cost for shipping (often not listed as part of the price). There is the delay and therefore a lack of immediate gratification after ordering a product. Or once a product is ordered, not all sites provide users with guaranteed delivery times or the ability to track shipments. For example, a company from which I recently purchased something online only promises delivery anywhere from 2 to 4 weeks with no facility for tracking the order once shipped.
The speed at the site, especially during the holiday season, could have been affected by network traffic. The fact that users now are less forgiving about technical shortcomings can understandably affect satisfaction.
All of these reasons are consistent with the suggested formula for an effective web site as implied by Forrester Research's Media Field Study for January. Their study implies that Content, Speed and Search are the three factors most important for a successful site. Content, especially effectively personalized content, fast sites, and effective search tools that allow users to find exactly what one wants, could all pertain to e-commerce sites.
At another level, these findings show that a well designed site should be well integrated with other facets of the business. For an e-commerce site this means integration with back-end systems such as inventory, accounting, shipping, etc. It's possible that some of the sites that hurriedly came on for the holiday season didn't have this and are suffering from it, unlike the more established sites such as Amazon.com. Just as data mining can be used to explore that reasons for satisfied customers it's also important that a data mining solution be well integrated with the facets of the business. In the early introduction of data mining technology to a business this is especially important - it must provide tangible benefits within the current practices and policies of the organization. This also means that data mining results are "actionable" i.e. discovered patterns can be easily translated into clear business activities and decisions.
---
For more information, see http://www.virtualgold.com.