Jakob Nielsen's Alertbox for September 1, 1997:

Why Advertising Doesn't Work on the Web

Because of the drastic differences in popularity between sites, only the top 0.01% of websites can generate sufficient revenues from advertising: in the larger picture, advertising is almost irrelevant for the success of the Web. Right now, Web advertising is attracting much media attention for two reasons:

Advertising is currently the only way for sites to generate a direct revenue stream (except for sites that support direct sales). This situation will change once we get micropayments (pay-per-view at the page level) as well as alternative revenue models like the link commissions paid by Amazon.com for people directed to their site by other sites.
Many old-media types only understand the mass-media business models typified by television. Maybe we can forgive advertising agencies who don't know any better, but an old-media perspective is also characteristic of many so-called "new-media analysts" who take an incredibly non-strategic view of the Web and analyze it purely in terms of "eyeballs" and a television metaphor (even though the telephone is a much better metaphor for the Web and its one-on-one and networking potential).

You Know Your Money is Wasted

There is an old saying in advertising that "I know half my advertising dollars are wasted - I just don't know which half!" Well, on the Web, you do know, since you can track how many site visitors come from which ads. As a matter of fact, you can continue tracking the users as some of them change from site tourists into paying customers. Only loyal users have any lasting value for your site, so you should refuse to pay for simple eyeball measures like CPM.

Web advertising should be valued in terms of the value of the business it creates from the new users it attracts to your site. This value is usually very small, which is why Web advertising works poorly and (while not completely useless) will be one of the smallest contributors to the future of the Web. Simple click-through is usually in the 1% range, meaning that 99% of the people seeing an ad don't even bother clicking on it. It is amazing how little most Web ads work at attracting clicks: they should recognize that they are one end of a hypertext link and that they have to create expectations as to the value of going to the other end of the link. This is called the rhetoric of departure in hypertext theory.

The other end of the ad's hypertext link is the pay-off page. Most often, these pages are highly disappointing and cause the user to back out immediately. This is why even click-through is a poor measure of the value of Web ads since it measures the alluring quality of your creative and not the ad's ability to deliver business.

Cognitive Medium

The Web is very different from television: it is mainly a cognitive medium, whereas TV is mainly an emotional medium. This makes TV much more suited for the traditional type of advertising which is flashy and promotes superficial qualities of products. While watching TV, people approach a vegetable state and the main goal of a commercial is to minimize interaction by keeping the user's hand off the remote control. As long as the user watches, you can keep them engaged by high production values and a message that says very little besides "we are good."

Where TV is warm, the Web is cold. It's is a user-driven experience, where the user is actively engaged in determining where to go next. The user is usually on the Web for a purpose and is not likely to be distracted from the goal by an advertisement (one of the main reasons click-through is so low). This active user engagement makes the Web more cognitive, since the user has to think about what hypertext links to click and how to navigate. This again makes the Web less suited for purely emotional advertising. The user is not on the Web to "get an experience" but to get something done. The Web is not simply a "customer-oriented" medium; it's a customer-dominated medium. The user owns the Back button. Get over it: there is no way of trapping users in an ad if they don't want it. (Actually, some new technologies do attempt to trap users, but I predict that such sites will be rejected by users. After all, there are plenty of customer-friendly sites to go to.)

The current slow download times work against emotional advertising. A pure branding message may work when embedded in the high production values of a TV commercial that can be viewed without any delay and without any action on the user's part. On the Web, everything is slow, and people don't like waiting for a fancy brand message.

Marketing That Works

Even though it does little good to run ads on other sites, there are many ways in which the Web can be used fruitfully in marketing. Most important are corporate websites where an entire site can be devoted to promote a company's products. Such sites should not be sales-driven but should focus on customer support and service, including detailed product specs and supporting information to facilitate the buying process. In other words: help customers buy; don't try a hard sell.

Banner ads are useful to the extent that they drive qualified users to such corporate sites, but there are many other ways of attracting traffic: a survey of people who had actually bought things on the Web discovered that only 12% of buying customers had arrived at the vendor's site from an advertisement - 88% of the shoppers had navigated there in other ways. Search engines and hypertext links are the most important mechanisms: offer content-rich pages, and other sites will link to you. You can encourage such linking by including appropriate URLs in your press releases and other PR efforts. Always include a URL in your print advertising (and remember to link to a pay-off page that follows up on the message in the ad; never link to your home page).

Finally, classified advertising works better on the Web than in printed newspapers, so I predict a successful future for Web-based classifieds: the one kind of ads that is a perfect match for the Web.

Comment added July 1998: Advertising still doesn't work on the Web and more people are starting to recognize this. UPSIDE Magazine stated that the clickthorugh rate is now 0.7%, that branding doesn't work on the Web, and that the only success stories are Ecommerce sites where people can buy stuff.

Comment added October 1998: The clickthrough rate has now dropped to 0.5% according to NetRatings: clickthrough has been cut in half during the year since I wrote this column.

September 15: Difference between intranet and Internet design

See Also: List of other Alertbox columns