The Design of Advertising Exchanges

R. Preston McAfee
Review of Industrial Organization

Add Remove

abstract Internet advertising exchanges possess three characteristics—fast delivery, low values, and automated systems—that influence market design. Automated learning systems induce the winner’s curse when several pricing types compete. Bidders frequently compete with different data, which induces randomization in equilibrium. Machine learning causes the value of information to leak across participants. Discrimination may be used to induce efficient exploration, although publishers (websites) may balk at participating. The creation of “learning accounts,” which divorce payments from receipts, may be used to internalize learning externalities. Under some learning mechanisms the learning account eventually shows a surplus. The solution is illustrated computationally. on 6/8/2019, 12:06:49 PM

excerpt In advertising auctions, three major features are: (i) the speed at which the auctions must be accomplished; (ii) the miniscule value and high volume of the items that are being traded; and (iii) the need to use automated systems for bidding, evaluation, and execution of the trades. 170 on 6/8/2019, 12:18:12 PM

excerpt The speed of display advertising auctions is breath-taking. After a user clicks on a link and a new page starts to load, the new page itself calls for an advertisement, known as an impression. That call for an advertisement spawns a call to an exchange to supply the advertisement. The exchange then holds an auction for the right to show that particular user an ad. The auction is run, the advertisement selected, pulled from a database, and then sent to the page, all in a fraction of a second. Speed is of the essence, because slowly loading pages create a bad user experience. Moreover, many pages won’t load the content until the ad loads, so that the user is left hanging until the ad is delivered. 170 on 6/8/2019, 12:18:55 PM

excerpt Not only is there no time to do complex things, the items are individually worth very little—usually less than a penny. Prices are sufficiently low that they are quoted in price per thousand, and suppliers of all but the most valuable audiences (e.g., medical and finance) would view $5 as a good price for a thousand impressions. In contrast to the old joke, however, a seller of advertising impressions can make it up in volume: The major exchanges trade billions of impressions per day. 170 on 6/8/2019, 12:20:00 PM

excerpt Automated systems are needed not just because of the speed of the auctions—a human wouldn’t be able to bid in a fraction of a second—but also because of the complex nature of the item that is being sold. The complexity arises because of the varied needs of advertisers. Some advertisers target demographics: age, gender, ethnicity, income, family status. Almost all have relevant geographic markets, ranging from a few ninedigit zip codes to a continent. Many advertisers target consumer interests such as sports cars, skiing, or Swiss cooking; and there are at least 5,000 distinct targeting variables. Further complicating matters are dozens of standard ad sizes. There may be restrictions put on the ads, such as no moving images, no skin except hands and face, or color limitations. 171 on 6/8/2019, 12:20:46 PM

excerpt For major advertisers, such restrictions create trillions of ad types, which could sell for different prices: e.g., a 200×300 pixel flash ad, shown to a 40–45 year old woman residing in Cambridge, MA, married, young children, family income $50–75K, interested in fashion, family cars, cross-country skiing, and books, currently visiting a news page about politics, appearing on Valentine’s day. Changing any one of those descriptors in principle might change the value to some of the advertisers, and hence the market price of the advertisement. Moreover, characteristics of the page and user are not the only relevant considerations: Advertisers have actually sought to advertise only in cities where the sun was currently shining, or only on days where the stock market was up over the previous close. 171 on 6/8/2019, 12:21:34 PM

excerpt Advertisers generally have access to three kinds of information about the opportunity to advertise. First, advertisers may have advertised to the user previously and written a “cookie” on the user’s computer. A cookie is a small text file that, in principle, is accessible only by the party that created the file. The contents of this cookie can be accessed by the advertiser to form a bid. Second, advertisers may have access to common information, such as the website that the user is currently visiting and the user’s IP address. IP addresses frequently provide the user’s geographic location, often within a kilometer. Since IP addresses are relatively stable, the IP address may be used as an index, and information about the user is recorded by the advertiser. That is, the advertiser can record all the previous pages in which it encountered that IP address, from which it is often possible to infer characteristics about the user. The current website usually provides a referrer (the previous page visited). Third, access to the information in the first two types of information can be purchased from third parties. 172 on 6/8/2019, 12:24:00 PM

excerpt Suppose that an advertiser bids $5 per thousand to advertise on a travel site, earning a reasonable rate of return. A rival may start bidding $6 for creditworthy individuals, thereby extracting most of the credit-worthy individuals and rendering the original $5 bid unprofitable because, say, the remaining impressions were of lower value. If the bidders have similar values for the advertising opportunities, data purchases are socially wasteful. Such cherry-picking drives bidders to increasingly refined bidding strategies, resulting ultimately in “real-time bidding,” where potentially every opportunity gets a distinct bid that is computed on the fly. 173 on 6/8/2019, 12:25:25 PM

excerpt The problem of matching advertisements and opportunities to advertise on web pages presents a remarkable opportunity to practice economics as an engineering discipline. The scale of the problem is unprecedented: arranging billions of transactions per day. The complexity of the problem is unprecedented: there are trillions of potentiallyrelevant product types. The speed of transactions is necessarily nearly instantaneous. The individual value of transactions is typically very small, requiring a “very low overhead” system. 183 on 6/8/2019, 12:28:32 PM

cites Internet Advertising and the Generalized Second Price Auction: Selling Billions of Dollars Worth of Keywords on 6/8/2019, 12:31:05 PM