The problem is, most advertisers do not have the size or the financial wherewithal of MasterCard simply to opt out of the automated system as it now exists. And even if they do, there are tantalizing financial incentives to stay put.
Much online advertising capitalizes on the lure of the so-called long tail of the internet — sites outside the biggest publishers and websites that draw relatively small but attractive audiences, like blogs for new parents or forums for truck enthusiasts. Advertising on those sites costs a fraction of what it does on more prominent online destinations, which typically deal directly with advertisers.
Money is funneled to smaller sites through a complex system of agencies and third-party networks, which can resemble a stock exchange. This system, known as programmatic advertising, allows brands to collect many millions of impressions — an industry term that generally indicates that an ad has been displayed and can be viewed.
But the lack of human oversight in this nascent industry has also led to confusion and mistakes. Technology has emerged to protect brands from showing up on sites that traffic in, say, pornography or spam, but those measures have been found wanting when it comes to disinformation. That kind of content is more difficult for security firms to detect than “open hate speech or nudity or violence or any of the normal stuff that you would stay away from,” said John Montgomery, executive vice president for brand safety at GroupM, part of the advertising giant WPP.
Joe Marchese, president of advertising products for the Fox Networks Group, said the system, set up to reward clicks and impressions, has fueled the growth of low-quality sites well beyond those focused on made-up political news.
“Honestly, the long tail is to advertising what subprime was to mortgages,” he said. “No one knows what’s in it, but it helps people believe that there is a mysterious tonnage of impressions that are really low cost. But low-cost impressions would mean low-cost human attention. How can any publisher of quality content survive on low-cost impressions?”
Marc Goldberg, chief executive of Trust Metrics, an ad safety vendor, said the effort to remove bad actors ignores that many advertisers value impressions over everything else. They would rather not choose and monitor what websites they are appearing on, he said, because they worry they will miss out on potentially lucrative destinations.
“What they’re doing is introducing all of these bad sites into our ecosystem and not having the means to monitor them appropriately and effectively,” he said. “The big problem in our industry is our expectations of scale are not aligned with reality.”
Relatively speaking, the money going to fake political news is minuscule, but it has taken on outsize significance because it reflects a world in which marketers are often not paying attention to where their ads show up.
Teenagers overseas and entrepreneurs in the United States discovered this year that they could earn thousands of dollars a month by writing wholly fictionalized or wildly exaggerated partisan political news intended to go viral on Facebook. They then reaped money from Google Ads and other networks after credulous readers in the United States clicked through to their sites.
“A lot of ad buying systems are trying to show the right ad to the right person at the right time, and you see that mantra of those three variables across the industry,” said Michael Tiffany, the chief executive and co-founder of White Ops, an ad fraud detection company. “Note how ‘on the right site’ doesn’t make the list.”
Workable, a software start-up based in Boston, recently joined companies like Kellogg and Allstate in pulling ads from the hard right website Breitbart and blaming automated ad technology for their appearance.
It is also watching out for its ads on fake news sites, which can slip through the cracks. Workable has about 110 employees and 6,000 customers but advertises on more than 50,000 sites a month through Google.
“Typically, Google has done a good job of keeping us off of low-quality sites,” said John Short, vice president for demand generation at Workable.
He added, “With fake news sites, or whatever you would like to call them, we need to be more alert.”
Facebook and Google have both said they plan to target misinformation, Facebook announced this month that it was experimenting with allowing users to report false stories and forging partnerships with outside fact-checking organizations.
This month, Robert Thomson, the chief executive of News Corporation, which owns The Wall Street Journal, described the digital ad market as “dysfunctional.”
Ad agencies, Mr. Thomson said at a conference held by UBS, should be “responsible for the audiences that they’re creating, and then what they’re serving those audiences and how they’re serving those clients, because we’ve sort of gone from the era of ‘Mad Men’ to mad metrics.”
Instead of using television networks, newspapers and magazines as proxies for reaching certain groups of people, many media buyers have been trained by automated advertising to use data and technology.
Andrew Casale, the chief executive of Index Exchange, an advertising exchange, estimated all the titles owned by the top 50 traditional media companies, including The Washington Post, Hearst and Condé Nast, probably account for 5 percent or less of the trillions of ad impressions available for sale each day.
Overall spending on automated banner displays and video ads will grow to $18.7 billion this year worldwide, and the United States will account for half of that, according to Magna Global, a major ad-buying firm.
“The problem, I think, for marketers is they’ve been trained to value media universally,” Mr. Casale said. “Fake news sites have really, really low prices and rates and tons of scale, so it’s almost like a drug.”
Mr. Tiffany of White Ops described one lure of automated buying: Say an advertiser wants to reach readers of The Atlantic or The New York Times who are in the market for a car. A relatively nondescript site might offer a price for 1,000 impressions that is one-fifteenth to one-twentieth as much as The Atlantic or The Times would charge, promising to deliver the same people who read The Atlantic or The Times, just later in the day.
Such transactions, occurring in real time en masse, open the door to fraudulent operators that use robots to impersonate people clicking on digital ads, he said, which can give buyers “a false expectation of the real achievable reach at a given price point.”
Mr. Thomson said he expected such programmatic advertising to come under even greater scrutiny.
“I would expect not only there’s an existential debate about news and fake news,” he said at the UBS conference, “but for that to flow into an existential debate about the role of advertising in society.”
Continue reading the main story