Automation Reveals Its Limits to Advertisers

Posted February 9, 2018 4:32 p.m. EST

For advertisers, one of the internet’s great promises has been the ability to automatically target people based on their interests and demographics, with little regard to the websites they are visiting. But these days, major brands have been forced to rethink how they advertise online.

Companies from Kellogg to AT&T have come under fire for inadvertently funding bigotry, hate speech and misinformation, often because they were using automated ad technology to reach groups of people across a vast number of sites and videos.

That’s a radical shift from the old way of doing business, when advertisers were far more dependent on individual outlets — say, The Atlantic or Good Housekeeping — for reaching certain audiences, using people and paper to place those ads. Now, advertisers of all sizes can essentially dispense their money into a complex system of agencies and third-party networks, which often resembles a stock exchange, using data to reach potential customers across millions of websites.

The perils of such systems came into stark relief last year with a new national spotlight on the spread of fake news and hate speech, as people began searching for the financial backers of such content. Consumers and journalists started visiting offensive YouTube channels and websites, taking screenshots of ads from prominent brands in those places and circulating the images on social media, demanding responses from companies. It was a new breed of customer service complaint; suddenly, one misplaced banner ad out of millions in a day could mean serious trouble.

An anonymously run Twitter account, Sleeping Giants, was a key driver of the effort. Now with 137,000 followers, it emerged in the weeks after the 2016 election and urged consumers to tweet screenshots of ads from brands like Warby Parker and Kellogg on Breitbart News, the alt-right site closely tied to President Donald Trump’s administration.

The account, which often paired its confrontations of advertisers with examples of offensive Breitbart headlines like “Birth Control Makes Women Unattractive and Crazy,” prompted many brands to blacklist the conservative news site and blame automated ad technology for their ad appearances. Other online movements bolstered such efforts, including #GrabYourWallet, which started as a campaign to boycott retailers carrying Trump-family products.

Brands were also discovered on sites promoting conspiracy theories, an especially worrisome issue after a man fired a rifle inside a Washington pizzeria because of false stories online tying the restaurant to child abuse. Allstate was found advertising on a site that claimed the Sandy Hook Elementary School shooting did not happen.

Screenshots and pressure from news outlets also helped spur a major brand exodus from YouTube last spring, after companies like AT&T and Johnson & Johnson were found advertising alongside videos that promoted offensive content, including promotion of terrorism. While YouTube defended itself by pointing to the large volume of content uploaded to the site every second, it has since introduced some changes that limit which videos can run ads.

Some major advertisers left last year with a new skepticism around the sky-high numbers of sites, videos and vendors required to effectively reach consumers. JPMorgan Chase made waves when it told The New York Times in March that it saw little change in the performance of its display ads after cutting back to 5,000 preapproved websites from roughly 400,000 sites a month.

The bank made the change after The Times discovered an ad for Chase’s private client services on a site called Hillary 4 Prison under a headline claiming that the actor Elijah Wood had revealed “the horrifying truth about the Satanic liberal perverts who run Hollywood.”