Can It?
But some experts say that much of today's E-mail automation software provides only a stopgap solution. The technique of matching keywords or patterns with possible answers for reps to cut and paste is one of the most popular, says David Daniels of Jupiter Media Metrix, "but that's never going to be efficient, because it's certainly not scalable." As volumes increase, he says, so must the number of customer-service reps. And companies that rely on canned responses may find themselves canned by customers: Daniels has found that fewer than 2 percent of E-mail autoresponse systems reply accurately to basic requests.
He is, however, more bullish on new products from Banter, YY Software, and others that tap artificial intelligence techniques such as natural language processing (NLP) and statistical modeling, and can spit out relevant responses up to 85 percent of the time. "Even with these tools, you're not going to take the human element out entirely," says Daniels, "but you can reduce your service staff and focus those who remain on your most valuable customers or the more complex problems."
Verizon Wireless cut the average cost of an E-mail inquiry in half, and Chad Kluko, executive director of E-service operations and development, hopes to drive it down further as Verizon's system expands to address such areas as billing information. The more important ROI measure for the company, though, is payroll. Because reps can now handle far more E-mail messages than before, the company recovered the cost of its Kana system within seven months; without the software, Kluko says, he would have had to add an astounding 3,000 staffers by 2005.
Leading CRM vendors have made E-mail automation a priority, with many either licensing or acquiring smarter artificial intelligence technology. San Mateo, California-based Siebel Systems Inc. is launching a product this summer that "solves all the problems every other E-mail automation tool has encountered," claims Matt Malden, head of service products. He says the new product, which incorporates both NLP and statistical modeling, can accurately classify 90 percent of messages, including negative ones like "Don't ever send me E-mails again." Messages like those often slip through the cracks of word-matching technology, because it misinterprets negatives, sarcasm, and common misspellings.
While the technology behind E-mail systems is one issue, the technology they run on is another. "Given the trend toward personalization," says Daniels, "these systems should ultimately be the hub for everything else, including the accounting system. In order to maximize that, you have to have a technology base that's open and extensible." Companies that have not made significant investments in a Web-oriented IT infrastructure face an uphill battle in getting all necessary systems talking to one another.
Security may also pose limitations. ABN Amro, for example, has technology in place that can automatically notify customers via E-mail, phone calls, or wireless messages every time a check is deposited in their account. However, "legal issues make it problematic to send sensitive financial information over the Internet in plain E-mails," says Matthias Autrata, senior vice president of the Netherlands-based company's Information Technology Services Co. "For that reason, we typically use E-mail only sparingly."
Chicago-based Allstate Insurance Co. faces a similar dilemma. In response to the CEO's directive to become more customer-centric, the insurer synchronized its customer databases across the organization and centralized many of its support functions at the corporate level, in an effort to get that "single view" of the customer that so many companies are pursuing. Allstate has been using software that relies on a "semantic processor" to "read" and classify incoming E-mail for the past four years, and has found it to be 95 percent accurate.
Given that the firm has so many of the pieces necessary for a fully automated response, and given that more than 80 percent of all incoming E-mails are requests for rate quotes, assistant vice president Rich Heneberry would like to move away from the current, time-consuming system of having agents approve each response before it goes out. But the risk of a botched computer-generated response has kept him cautious. "Our attorneys always remind us that we're legally responsible for whatever we send via E-mail, so we're always looking at that trade-off between efficiency and risk," he says.
Mail Bonding
In theory, handling inbound E-mail well should enable a company to make the most of the marketing potential in outbound E-mail. Customer service data from those E-mailed complaints and questions, for example, should wind up in a company's marketing database, providing plenty of detail for highly targeted, personalized E-mail campaigns. "If you're using E-mail as a mass medium, you're missing the boat," says Rogers of Pepper and Rogers.
Done right, such campaigns are more effective than any other media in driving online purchases, according to 50 marketers recently surveyed by Forrester. When Los Angelesbased Ticketmaster Corp. targeted its E-mail ads for Bruce Springsteen's most recent tour based on users' buying history and geography, it achieved a 47 percent response rate and a 20 percent buy rate, compared with an average 1 percent response on previous campaigns.
Reader Comments» Post a comment