It's 9 p.m. Do You Know Where Your Ads Are?
This op-ed first ran on Forbes.com
By George John, the CEO of Rocket Fuel.
“Change is good.” So goes the tagline of arguably the first viral commercial – a Doritos ad featuring recently defeated governors Ann Richards of Texas and Mario Cuomo of New York talking about “change” as they munched Doritos from the newly changed and re-branded packaging.
Change has come to the $30 billion digital advertising industry. The “Mad Men” days are over (well, the suits and secretaries are gone, but you could argue drinking, ennui, and client resentment are still fixtures of Madison Avenue). In the old days, agencies used to place an ad in relatively few places. “Give me the back cover of Life magazine and a TV spot at the beginning of Mutual of Omaha’s Wild Kingdom,” a media director could say, and then the agency and client could just buy the magazine and turn on the TV to see that the ad was correctly placed.
If you could teleport “Mad Men”’s media buyer Harry Crane directly from 1968 to 2011 and ask him to plan a campaign, he wouldn’t believe we now talk in terms like online ad networks, real-time bidding, inventory exchanges, data exchanges, and offline metrics studies.
But there is one parallel between the life of a media buyer in the 1960s and today: Media buyers today still mostly measure success via simple exposure. Spend $1 million to reach 10 million consumers – good. Spend $1 million and reach 20 million consumers – better!
But now sophisticated marketers are in a new quandary. They need to show not just that the ad was shown to a large number of people, but that it actually had some effect on a business metric that the company and its investors care about. This has led to the biggest change in marketing practices in the last 40 years: a shift toward rational, computational, and algorithmic media buying.
An algorithm is a recipe for computation. Algorithms tell your GPS in your car how to figure out the best route to your destination, and they power adaptive medical devices such as the advanced pacemakers that keep Dick Cheney alive. In the 1960s, while Don Draper was making a fortune at Sterling Cooper, John McCarthy, Marvin Minsky, Nils Nilsson and other visionaries of Artificial Intelligence at Stanford and MIT were imagining a world where computers could think faster and better than people.
If you let AI algorithms replace Harry Crane, you don’t buy the back cover a magazine seen by 20 million people, but rather the algorithms buy 20 million specific digital ad placements to reach 20 million different people, segments, times of day, cities, browsers, and millions of other audience metrics. Machines can do this today with real-time bidding.
But to find out if your 20 million ads are working is the hard part. In an algorithm-driven media buying world you can’t just buy Life magazine and look for your ad on the back cover. Instead you have to understand why you were running the ad in the first place and measure the real impact of the campaign. What is the goal you were trying to achieve? Is it awareness of a new product? Brand favorability over your competitor? A lift in sales? Driving social connections? Among which specific audience segments?
The policing job – making sure ads don’t turn up on questionable sites – is also more complicated today. You don’t just have one placement to police, you have millions. Unsurprisingly, the answer here is to use yet more algorithms to automatically score all placements for brand safety.
Change is good, and the mental energy spent figuring out how to manage algorithmic campaigns pays off in results that are five times better than before. Better ads are good for the economy (more efficient ads motivate consumers to buy products); good for advertisers (who can channel wasted advertising spend into product innovations); good for media companies (who can program their content with less concern for the demands of one particular advertiser, since thousands will buy their ad inventory); and good for consumers (who are less annoyed by irrelevant ads).
Media buyers shouldn’t be afraid of algorithmic buying; it won’t put them out of a job. Instead, they just need to become technology mavens who understand that buying ads today doesn’t happen via a phone call to Life magazine.