In the last two years, real-time bidding (RTB) through online advertising exchanges has secured a place in the future of digital advertising. Some of the major technology companies have acquired ad tech start-ups to create their own solutions for the various players in the online advertising market. Yahoo bought Right Media, Google bought DoubleClick, and Microsoft bought AdECN. Examples of ad exchanges that have not been acquired include AppNexus, PubMatic and adBrite. According to Brian O’Kelley, CEO and co-founder of AppNexus, RTB on ad exchanges began in early 2009 and went from selling a few million ads online to a billion by the end of 2009 and 8 billion by the end of 2010. The advantage of RTB on ad exchanges is that it provides ad networks with a way to target each individual user (user-based targeting) to take advantage of the rich user data collected by different ad networks, publishers and others. And, because RTB means a live auction for each ad from a participating publisher/ad network, these players in the ad market only pay what they are willing to for each ad served to an individual user.
I got to interview a friend who works at one of the above mentioned online ad exchanges about the actual process that occurs behind the scenes of these exchanges. He asked that he remain anonymous and that I don’t directly identify the company. So then, Joe-Bob works for Addition ad exchange (names made up). The following will hopefully give you a rough idea of how an ad exchange works in real life:
The process begins with a company (such as IKEA) hiring an ad network (like Collective) to create an online advertising campaign for them. Together they craft a strategy for whom to target–like new college students, people looking for a new apartment, or people browsing on other furniture stores’ websites–and how much to pay for each of these targets (per impression and per click). This strategy is used to create an algorithm for bidding on ads. Then the ad network comes to the ad exchange. Joe-Bob’s ad exchange simply hosts the auction for each ad served online. In the milliseconds between the time you hit enter after typing in a URL or click on a link and the loading of the website, any publisher whose own ad network uses the ad exchange will auction off each of its banner-ad spaces. If a user (me) who lives in New Haven, CT is looking at apartments online in New York City, then IKEA might want to bid on the ad-space on whatever website that user goes to next. But the user (me) might also have recently read an article on the effectiveness of zinc lozenges in fighting colds (GUYS, this stuff works SO well! I got rid of my last two colds in three days!). Cold-EEZE might want to compete against IKEA for an ad served up to the user (still me). Each bidder has an algorithm that battles all the other bidders’ algorithms in a competition for the ad space (without algorithms doing the bidding, the auction would take far too long). The highest paying bidder with an ad that the publisher accepts wins the auction and gets their ad served.
All ads and ad-spaces are coded/tagged to ensure that an ad doesn’t end up in the wrong place. For example, Disney does not want violent, sexual, etc. ads on its website no matter how much the ad network wants to pay for the space. Any ad that does not follow the guidelines for advertising on Disney’s websites will be disqualified from the auction. Users are also tagged. Each user is identified only by a long (20 digit?) number (your true identity is not sought after by the more “ethical” user data collectors). Joe-Bob showed me my own online profile. Much like the “issue publics” discussed in Phillip Howard’s book New Media Campaigns and the Managed Citizen, each user is coded into as many different categories that he/she fits into. In the earlier example, I discuss the ads I might get served from being coded for moving to NYC and for reading about cold remedies. These categories that I fell into in the example give advertisers the power to target ads that I might actually find relevant.
The issue of data privacy is a legitimate one. Joe-Bob explained to me that his ad network does not collect this data itself, but allows the participants in the auction to use their own data (however they obtained it). Most user data is obtained by using cookies (the kind in your web browser, not the kind you eat). According to Joe-Bob, standard cookies expire after two weeks, and users’ online IDs change frequently. This means that data about you is not usually being compiled over years of browsing/buying history to hone in on what will make you freak-out and immediately buy anything you see in an ad. Instead, if they track a batch of user IDs over a period of time, after about two weeks the user data for those online IDs will no longer be accumulating and the batch will be useless.
After only two years, the results of real-time bidding (RTB) on online ad exchanges is phenomenal for everyone in the online advertising business. Publishers on ad networks using RTB ad exchanges have seen ad prices jump. Google claims that publishers sold ads for 130% more ON AVERAGE using DoubleClick ad exchange than the ads sold using older methods. Other ad exchanges have also recorded significantly higher prices for ads. Duh! Targeted ads are obviously worth more, and the same data collection gives advertisers the ability to track user activity long after an ad-impression or click (if I buy IKEA online next week, they can give credit to the ad I saw today).
With newspapers and other online content creators struggling to make money, the growing use of RTB ad exchanges powered by user-data targeting is part of the bright light at the end of the tunnel. With more money flowing through the online advertising system, everyone wins (well, except for people freaked out by targeted ads and user data collection).