1 2 3 4 5 6 7

measuring Internet use. When Hoffman and Novak set out in mid-1996 to survey Web-based advertising, they discovered there were more than ninety sizes of Web ad banners, little consistency among sites in pricing advertising, and little demographic information about customers. Sites don't even necessarily charge the same way. While most charge by the thousand "impressions"--the number of times an advertising banner is displayed on a page downloaded by a visitor to the site-- some charge by "clickthrough," that is, the number of times a visitor clicks on an advertising banner to go to the advertiser's own site, the disadvantage being that the sponsored site loses out if the advertiser's design is too poor to attract those clickthroughs.

Differences and inconsistencies like these led Hoffman and Novak to call, in September 1996, for the development of standardized techniques for measuring Web use. "First, there are no established principles for measuring traffic on commercial Web sites that seek to generate revenues from advertising sponsorship," they wrote. "Second, there is no standard way of measuring consumer response to advertisements. Third, there are no standards for optimal media pricing models. Finally, the complexity of the medium in general hinders the standardization process."[12] Nonetheless, they note, estimates are that Web-based advertising will grow substantially over the next few years, and predictions put advertising revenue for the year 2000 at $1.7 billion to $5 billion.[13] To put those numbers in perspective, in 1995 U.S. companies spent $31.2 billion on direct mail and $11.1 billion on radio advertising. In 1996, Web advertising was a tiny fraction of those, and two-thirds of that went to the top ten sites.[14]

This is one of the reasons that predictions that large corporations will fare badly on the Web will probably be proved incorrect. The Web's lack of familiarity and raw immaturity as a medium reward familiar names disproportionately. Buying groceries through one of the few trial systems available at the end of 1996 meant choosing products from a list offering little auxiliary information; the inevitable result is to gravitate toward national brand names and known quantities.

In order for Web sites to have a shot at turning commercial--sites like the Internet Movie Database and the search engine Yahoo! started as personal projects--they have to collect demographic information to show potential advertisers that they attract the kind of visitors those advertisers are interested in. The simplest way of collecting this information is the registration forms required by many sites, especially those publishing new, professionally written material--what they call "content."

Unfortunately for site owners who are trying to find a way to pay themselves a living, this kind of information isn't exactly what a lot of Net users want to supply. Attempts to log user characteristics are seen by many Netheads as an unwelcome invasion of privacy; in fact, this was named as the most important issue facing the Net by 26.2 percent of respondents to the October 1996 GVU survey of user attitudes. Besides, people hate looking at ads; that's why we now have VCRs that mark the ads and automatically spin past them. The Net equivalent is services like the Anonymizer and Internet Fast Forward,[15] which allow anonymous browsing, and intermediary services that allow you to register once and then browse using a single, numbered ID. The third-party service makes only the aggregated demographic data available to its subscribing sites, while users only reveal their personal details to one centralized site.

Many Web wanderers don't realize just how much information Web sites can acquire about them while they're visiting, even without registration forms. Most of this is pretty harmless--do you really care if the Alien Abduction Web site marks down that you're using Windows 95 and Netscape 3.0 while it's asking you to sign up for a date and time to be abducted?[16] But it spooks people to think that a Web server can read information like that off what they think of as their private, personal computer systems. Web sites can also typically tell what domain you're coming from (though generally not your precise email address) and what pages you look at and when.