There has been a lot of talk recently about the failure of the World Wide Web as an advertising medium. Even the best-known sites, such as Time Warner's Pathfinder, are not generating ad revenue. Advertisers perceive that online ads do not achieve returns on investment equal to those of other media, and banner ads displayed on Web pages produce only a small number of sales.

But in the rush to cyberspace, many advertisers seem to have missed a crucial point: online ads function differently from the way television or magazine ads operate. In the battle to secure real estate on the electronic frontier, there was no time for defining how the process of reaching the online buyer could best be achieved, considering the properties of the medium.

To understand why banner ads are failing, we should define what they are attempting to achieve. Is that goal appropriate? How can they maximise returns on cyber-investment for the second generation of Web advertising?

Conventional advertising aims to attract the attention of potential buyers by using imagery and words. A television ad for a car tries to be eye- catching and tempting enough to seduce the would-be buyer into the showroom. At that point, the buyer can obtain more information about the car, and perhaps take it for a test drive. Potential buyers will also be exposed to sweet-talking sales staff trained in consumer psychology.

This process looks very different online. If I click on the banner ad for a new piece of software, book or CD, I can immediately obtain information that will help me to make a decision. Then, if convinced, I can download a demo and sample the product. If I like it, I can enter my credit card number and download a full copy of the product (in the case of software) or order mail delivery (in case of a book or CD). So an ad on a Web site not only grabs attention, it also acts as point of information, a point of sampling and ultimately as the point of sale - all at one tiny, virtual spot.

As a buyer, I'm in control of' the process, avoiding sweet-talking sales people and left in peace to make a choice based on an understanding of the product, its advantages (or lack of) over competition, and its price. Given full data, I will be more discerning and focused on the quality of the product itself, as opposed to its image or pretty graphics on a sales brochure.

The ability to provide relevant product information will become the Web advertiser's main weapon in the battle for sales, rather than having the most showrooms in town, the biggest TV advertising budget or the most persuasive sales force.

We are witnessing a rapid re-engineering of the process of sales, with key variables (distribution channels, marketing budgets) losing their importance to old-fashioned features such as quality and price.

How are we going to improve Web advertising to support this process? The war of the Web sites must make use of smart information management to improve delivery of information to the user. This has been done so far through the use of intelligent agent technology, which determines individual characteristics and preferences of the potential customer, matches them against the products available and then informs potential customers about specific products or services.

It puzzles me why so few Web advertisers have adopted this technology. For example, a trip to the Computershop Web site is a major hassle, as it requires me to sieve through all sorts of dross about millions of new products, while all I want is to know whether the new version of my modem has been shipped yet. Even better, if I were known to my modem's manufacturer, they could remind me to visit their Web site when a new product arrives.

It would help me a lot if I could enter the key information about my computer, configuration and peripherals and then let the intelligent agent do the thinking and let me know when a new product is available. I would be much more likely to test new products and buy them online, knowing that somebody up there is monitoring my needs, saving me online time and telephone bills.

This use of precision information by Web advertisers will increase the ratio of buyers to viewers and make banner ads work harder.

Meanwhile, I'm off to try to sieve through imprecise and non-personalised Web-based information on the site to book a break that will not only match my preferences for hot weather and great food, but also help me to avoid arguments with my friend, who would prefer extreme sports in the Antarctic circle. The first travel Web site and intelligent agent that will sort my problem will buy my gratitude and custom for the next decade. Relevant URLs can be sent to

Eva Pascoe is co-founder of Cyberia Internet cafes. She will be writing a fortnightly column for the Network section.