InformationWeek.com

Welcome Guest. | Log In| Register | Membership Benefits

March 20, 2000

Companies See Gold In Outside Data Analysis
Customer-relationship management and the Internet are fueling the growth of outsourcing

By Barbara DePompa

Related links:

Outsourcing Adds Complexity (1/10/00)

TechEncyclopedia

Need a definition of a technology term? Look it up here:

Send Us Your Feedback

t first glance, it appears illogical: giving up control over critical business information by outsourcing a data warehouse. But some companies are finding it's easier to focus on developing, manufacturing, and selling their core products and services if they leave the data analysis to the experts.

What's driving this change? It's the Internet, and the big shift that's occurring as customer-relationship management--the "true real-world" application for data warehousing, according to Scott Nelson, VP and research director at Gartner Group--takes hold in all kinds of businesses. "Why gather data if you can't make money with it by using it to improve your marketing strategy?" Nelson asks.

Jewelry.com is doing just that. The Internet jewelry retailer went live in November and has handed over its clickstream data to Candle Corp. for warehousing, data mining, and other types of analysis. Jewelry.com is using Candle's CandleNet E-Business Assurance Network (EBAN) service, which provides application service monitoring, collects response-time data, scrubs data for warehousing and associated reporting, and performs online analytical processing data analysis. Users can then view the analyzed data via a Web browser at the EBAN Portal.

Based on clickstream analysis of visitors to Jewelry.com in its first month of operation, the site clocked 15,000 visitors per day, with 90% not going past the home page because the site was too slow. Jewelry.com quickly boosted performance by adding Internet service provider Digital Island Inc.'s Sandpiper 500 caching servers, and "suddenly 10 times as many visitors were going deeper into the site," increasing sales, says Paul Rajewski, chairman and chief technology officer at the online jewelry retailer. The El Segundo, Calif., company had a good Christmas season--more than 1,200 customers bought jewelry from the Web site, and Candle's EBAN system kept track of key performance metrics so customers weren't bogged down by slow performance.

According to Rajewski, speedy performance is key to keeping online customers happy. "Expect to spend a minimum of $200,000 for an E-commerce system," he says. "We were able to set up our Web site in six weeks, but we've already upgraded systems twice to reduce bottlenecks and speed performance."

Jewelry.com's current configuration includes about 20 Sun Microsystems processors and 40 Gbytes of RAM, line speed of 45 Mbps, and the Digital Island caching servers, which take a huge processing load off the Sun servers. Jewelry.com uses Intershop's shrink-wrapped E-commerce software package, which helped speed the site's development.

Photo by Edward Carreon The fledgling E-retailer is starting to advertise its online wares, register site visitors, and offer special promotions to entice repeat visits. For Valentine's Day, Rajewski placed ads aimed at male readers in Sports Illustrated and The New York Times to analyze the percentage of male vs. female buyers who shop on the site. The ads offered a discount and required shoppers to fill in basic demographic information. "We'll use the analyzed results to help formulate our media advertising plans for the coming year," he says.

"Businesses need to analyze what's happening on their Web sites and use tools to investigate those Web logs or find outside expertise to help," says Richard Rist, president and founder of the Intellor Group, a consulting firm that specializes in data warehousing and analysis. Clickstream analysis is a natural for outsourcing, he adds.

E-businesses are uniquely designed to enable outsourced data warehousing and analysis, because their data is gathered, tracked, and stored on the Web. Most traditional businesses store customer information in a variety of computer systems. This makes it more difficult to gather, sort, and accurately analyze customer profile information.

continued...page 2, 3