Welcome Guest. | Log In| Register | Membership Benefits

News In Review

March 15, 1999

Print this story
Print this story
Tapping The Pipeline

Web sites can offer a wealth of customer data; smart companies are mining, analyzing, and acting on it for competitive advantage

By Clinton Wilder

Related links from our sister publications:
  • Planet IT Planet IT E-Commerce Technology Center

  • Computer Reseller News Microsoft Makes Big Pitch For E-Commerce For Small Businesses

  • T he first wave of Web development is over. Virtually all organizations have Web sites, and most of their customers are comfortable doing business online. Now comes the hard part--turning those Web sites into a channel for sales, customer service, and information gathering. That means making sense of the signature of a successful Web strategy: the torrent of information flowing in via the Internet.

    The numbers are staggering because every customer action on a Web site, from a simple click-through to a complex buying transaction or product configuration, generates data that can be captured and mined. The most popular sites handle more than 100 million page views a day. Barnesandnoble.com receives more than 2 million unique visitors each month. Realtor.com processes more than 250,000 home searches a day. And the numbers are growing--at busy times, Amazon.com receives 400% more orders than it did a year ago.

    Angie SnellingPhoto by Gordon Morioka Savvy companies aren't just dealing with this flood of data, they're embracing it. For companies that can figure it out, the Web represents a direct pipeline to customer behavior, taste, and opinion--the value of which can't be approached by market research reports, direct-mail response rates, or focus groups. "We like data," says Angie Snelling, director of electronic commerce for machine tools manufacturer Milacron Inc. "We want to know everything."

    Figuring it out isn't easy. There's the ever-present privacy issue. Most companies insist that they use Web data to analyze behavior by groups, not individuals. Products to help collate and categorize Web data are immature or scattered across several categories. And the expertise to fully exploit Web data in most companies cuts across several business areas.

    "As sites move from driving traffic to really focusing on return on investment, that makes a whole different kind of tracking and data mining necessary," says Pyramith Liu, director of sales/company stores for PC maker Acer America Corp. in San Jose, Calif. Acer is building tools that will track online customer behavior inside its Web site's firewall. That data will help the company understand, for example, what a customer does after buying a product, rather than just which banner ad the customer clicked on or which link he or she followed to the site, says Liu.

    Acer is also working with Open Market Inc., its commerce server vendor, to develop better data collection and analysis tools for Open Market's Transact server and LiveCommerce electronic-catalog products. Open Market has just finished defining requirements and is starting the development process for a data collection and management module for those products, with delivery planned for the third quarter.

    "Log-file analysis is not the answer," says Paul Baier, director of enterprise marketing at Open Market. Instead of long lists of raw numbers, business users need data they can act on, says Baier. For example, marketers need to be able to interpret the interplay between promotions and customer orders. "The key is getting the data out in a nontechnical format," Baier says.

    Another key, say analysts, is focus. "The companies doing this effectively don't collect all that much information," says Steven Johnson, co-director of the E-commerce program at Andersen Consulting. "It's not so much collecting it, but choosing to act in response to it."

    Milacron launched its business-to-business Web site two months ago to sell its machine tools online. The Cincinnati company is analyzing the online technical help its customers seek in a section called The Wizard, and routing that intelligence to its research and development engineers. "For the first time, we can start to make R&D decisions based on actual data--X many people say they want this type of product improvement," says Snelling.

    Direct Links
    For companies such as Milacron that sell mainly through distributors or other third-party channels, the Internet represents the most direct information link they've ever had to their ultimate end customers. "The Web opens us up to people we never talked to before," says Phil Gibson, director of interactive marketing at National Semiconductor Corp., which posted one of the first searchable electronic catalogs on the Net in 1996. "We used to get feedback, but it was always third- or fourth-party. Now it's direct and immediate. It comes in off the Web site, right from the design engineer's standpoint."

    National Semiconductor divides the Web data it captures into "digital" and "analog." Digital refers to Web-site usage statistics; the company uses Accrue Software Inc.'s Accrue Insight to capture navigational data, such as how many clicks it takes a user to reach specific areas of the site. Based on that data, National Semiconductor reduced the average number of screens a user must click through from seven to two, says Gibson.

    Analog information consists of customer E-mail messages, and National Semiconductor employs technology to analyze those, too. The company wrote Lotus Notes-based automatic filtering programs that take E-mail messages from a Domino server and route them to the appropriate product manager. "Every product-line marketing person gets a report on how their product is doing online," says Gibson. "The click-through activity from Accrue, the E-mails, and the sales forecasts from our resellers are all imported into Notes, and those three together give you a pretty good picture." Based partially on Web-site feedback, National Semiconductor has expanded its product line of temperature sensor chips from three devices generating less than $10 million in annual sales to some 25 products generating more than $100 million in revenue. The chipmaker also used Web data to support the decision to phase out some 3,000 low-selling chips from its product line of 8,000 devices.

    Third-Party Help
    Partly because analyzing Web data is so difficult, some companies use a third-party service to collect the data and help put it to marketing and business use. The New York Times Co. uses IBM's SurfAid Analytics, a combination of data mining products and services, to help determine how users are interacting with the publishing company's Web site.

    continued...page 2, 3

    Read sidebar story, "Bear Stearns: An Online ".edu"-cation."


    Photo of Snelling by Gordon Morioka


    Back to This Week's Issue

    Send Us Your Feedback

    Top of the Page