The Internet measurement and research firm is rolling out a product that measures activity on individual sites, an area of online usage research once considered more or less off-limits to companies that measured traffic throughout the Web. "The two services are complementary because they are measuring two different things," said Seth Segel, senior VP at Jupiter, which has struggled amid the Internet downturn (its stock market capitalization last week was $15 million compared to the $360 million of rival NetRatings).
In short, what the company is undertaking with its new product is a more efficient way of tracking traffic on individual sites, replacing the reams of log files many sites shift through to determine their usage. Customers will install a "400-byte, standard piece of HTML [code]," Mr. Segel said, that will effectively measure user activity deep within a site.
"We're measuring what a user sees in the browser, not what a server sends out," he said. The site measurement service will be able to track the accessing of cached pages, essentially copies of Web sites that users sometimes access through another service such as their ISP. Jupiter would not identify what clients are involved in the launch. Pricing will vary according to client need.
As the company rolls out its site measurement tool, it enters territory that seems bound to start controversies anew over which measurements are most accurate. While the existing Web measurement service monitors a panel of about 100,000 people over the entire Internet, the site measurement business will likely report different numbers. But one Internet media executive said that shouldn't present a problem. "I think the market has been sort of educated to that," said Scott Moore, publisher of Microsoft Corp.'s e-zine Slate.
Mr. Moore, who said Microsoft hasn't decided whether to use the new Jupiter product, feels there is room in the marketplace, and even within one company, for different measurement results. He also believes housing both types of research within one company could help to rectify some discrepancies in measurement, which often show panel traffic seeming to undercount compared to individual site data.