Chapter 3 of 11 · 8 min read
Have You Ever Clicked Here?
From The Market for Your Mind by Wrotebook
The advertisement was a strip of colour a few hundred pixels wide with a sentence on it, and the sentence was about the reader's hand. "Have you ever clicked your mouse right HERE?" it asked, with an arrow pointing to the answer, which was printed in capitals: "YOU WILL." It appeared on 27 October 1994 at the top of pages on HotWired, the new online offshoot of Wired magazine, and it was paid for by AT&T, which at the time was running a series of television commercials under the same slogan, predicting with great confidence the marvels of the near future. The banner was the slogan applied to itself. It did not sell a telephone line or a calling plan. It sold the act of clicking on it, and it was right.
AT&T was one of about a dozen advertisers that bought space on HotWired that autumn, and a good deal of the early web regarded the whole transaction as faintly vulgar, rather as a village might regard the arrival of a betting shop. Whether the AT&T strip was the first paid banner on the web is disputed, since several went up at once and the alcoholic drink Zima has its own claim, and the dispute has the agreeable quality of most arguments about firsts, which is that it matters a great deal to the people who were there and not at all to anyone else. What made the AT&T banner famous was a number. Its makers later said that 44 per cent of the people who saw it clicked on it. The figure has been repeated ever since, almost always without the observation that it comes from the people who made the advertisement, which is the kind of source that tends to be pleased with its results.
Suppose the figure is roughly right. It would not mean that the advertisement was persuasive. It would mean that clicking on things was new. In 1994 the web was a novelty in nearly every particular: that a page could live on a computer in California and appear on a screen in Bristol, that a picture could arrive line by line like a blind being raised, that a word could be blue and underlined and take you somewhere else when you touched it. An invitation to click was, briefly, an invitation to do the most interesting thing a computer could do. The audience was small, young, curious and unusually well paid, and it had not yet learned to look away.
It learned with remarkable speed. As more sites sold space and more advertisers bought it, banners multiplied, and as they multiplied they competed in the only ways available to a rectangle: they grew, they flashed, they wobbled, they announced that the reader had won something, which the reader had not. The proportion of people who clicked fell year after year until clicking on a banner became something most people did chiefly by accident. Researchers who tracked where people's eyes went on a web page found that many simply stopped looking at the parts of the screen where advertisements usually sat, as though those regions had been painted out. The condition was named banner blindness, which is a gentle way of saying that an industry had trained its audience, at considerable expense, to see through it.
The banners were, for all that, paying for things, and the things were real. A great deal of the early web was given away: news from papers on the other side of the world, email accounts that cost nothing to open, free space on which anyone with a modem could publish a page about their dog, their poetry or their theory of the Kennedy assassination. A teenager in a small town could read the same journalism as a professor in a capital city, and the price was a strip of colour at the top of the screen that they had already stopped seeing. Somebody had to pay the people who wrote the journalism and the companies that kept the servers running, and the advertisers did, on the understanding that they were buying attention. Whether they were getting it was a question the arrangement worked hard to postpone.
The industry's answer to people who had stopped looking was to make the advertisement harder to avoid, and the purest expression of that idea was the pop-up, an advertisement that opened in a window of its own on top of the page you were trying to read. Ethan Zuckerman, who worked at the web-hosting company Tripod in the late 1990s, says that he wrote the code that launched it, and in August 2014 he wrote an essay in The Atlantic apologising, as part of a larger argument that advertising had become the web's original sin. By his account the idea had been to keep an advertiser's message from appearing to endorse whatever a user had put on their own page, which is a reasonable thing to want and an unfortunate thing to have achieved. The pop-up became the most hated format on the web until browsers began blocking it, at which point the advertising moved elsewhere and the arms race resumed on new ground. Interruption had met inattention, inattention had won the first round, and interruption had come back with a bigger window.
The pop-up would have been a footnote if the web had been only a new place to put advertisements. What made it different, and what made the people buying advertising so excited about it, was that it could count. Television had to estimate its audience from a small panel of homes, a sheaf of diaries and an agreement between the parties to believe the result. A web page could count everything. Each time a page was sent to a screen with an advertisement on it, a server could log it as an impression. Each time someone clicked, the server could log that too. It became possible, for the first time in the history of advertising, to say exactly how many times a message had been shown and exactly how many people had acted on it, to the unit.
The click quickly became the web's currency, and it was a currency with a much better reputation than the rating. A rating had to be taken on trust; a click left a trace that could, in principle, be inspected. Advertisers who had spent decades paying for audiences they could not see began paying for actions they could, sometimes by the thousand impressions and increasingly by the click, and the ones who paid by the click felt, with some justice, that they had finally stopped paying for people who were in the kitchen. The new currency had its own small comedy. A publisher's server and an advertiser's server, counting the same campaign, almost never arrived at the same figure, and a surprising amount of the early industry's time went on arguing about whose log was right. The argument was never entirely settled, and it was eventually resolved in the traditional way, by agreeing which count everyone would pay on.
Imagine, then, the invoice HotWired might have sent AT&T for that first banner, if invoices were obliged to tell the whole truth. It might have read:
Supplied: one strip at the top of a page, read by early adopters who had never seen anything like it.
Paid out to readers: a novel thing to click, and a page at the end of the click.
Taken in from readers: a moment of curiosity, one movement of the hand, and their arrival on a page of the advertiser's choosing.
Measured: impressions served and clicks recorded, exactly.
Not measured: whether anyone remembered the advertisement the next day, thought better of the company, or ever spent a cent with it as a result.
The last line was the one the web promised to fill in. For a century advertisers had repeated a complaint usually attributed to the American retailer John Wanamaker, who may or may not have said it, that half the money they spent on advertising was wasted and the trouble was that they did not know which half. The web would tell them. Every pound could be traced to an impression, every impression to a click, every click to a visit and, eventually, to a sale. Marketing would become a science, and waste would become a line in a spreadsheet that somebody could be asked to reduce. The people who sold the space would provide the numbers, and the people who bought it would finally be able to check.
Part of what made that promise plausible was a small piece of technology devised in the same year as the banner. In 1994 Lou Montulli, a programmer at Netscape, which made the most popular web browser of the day, came up with a way for a website to leave a tiny file on a visitor's computer, so that the next time the visitor came back the site would recognise them. He called it a cookie, after a programmers' term for a token passed between programs, and its purpose was entirely domestic. The web had no memory; every visit to a page was, as far as the page was concerned, a first meeting. The cookie let an online shop remember what you had put in your basket while you went to look at something else, which is not a sinister thing to want from a shop.
The trouble with a memory is that it is useful to more than one party. A site could set a cookie for its own visitors, and so could anything embedded in the site's pages, including the advertisements, and the companies that served those advertisements were present on a great many sites at once. If the same advertising company placed its cookie on your computer from a hundred different pages, it could recognise you on each of them, and begin to piece together where you had been and what you had lingered over. Nobody had to sit behind a mirror any more, or pay eight women in cash and coffee. The notes were kept automatically, across millions of people, on servers none of them would ever see, by companies most of them had never heard of.
What the web delivered, in place of the end of waste, was precise measurement of something smaller than attention. A click is not a person thinking about a product. It is a finger moving, for any of a dozen reasons, among them curiosity, confusion, the promise of something free and a crowded page on which the advertisement had been placed exactly where a thumb was likely to land. An impression is not a person seeing an advertisement. It is a server recording that an advertisement was sent to a screen, which might have been looking elsewhere, or been scrolled past in a fraction of a second, or been sitting in a tab nobody opened. The numbers were exact, and what they were exact about was narrow, and before long the market was paying for the narrow thing with great precision. Publishers who were paid by the click put their advertisements where clumsy hands would find them. Publishers who were paid by the impression found reasons to split an article across several pages. It was the old arrangement with more decimal places: an agreed count, standing in for a thing nobody could count, with everyone paying on the count.
The cookie, meanwhile, did its work. It had been designed so that a shop would remember what you meant to buy. It became the means by which a great many companies that were not shops, and from which you had never meant to buy anything, would remember what you had looked at and not bought, which turned out to be a much larger market than the basket, and one in which the person being remembered was the only party not being paid.