Abstract
When The Times set out to expose dark practices in digital advertising the trail led to the web giants
It was not, in all truth, a great tip-off. In September 2016 I was told that media agencies, the obscure companies that buy advertising space on behalf of brands, indulged in certain murky business practices.
Being a reporter not focused on business stories, my initial reaction was muted. Even if the story proved to be true, even if sources could be found and documents obtained, even if the whole thing worked out perfectly, would such a story make the front page? I doubted it. The Times excels in its business coverage but it is fundamentally a general-interest newspaper. However, since I didn't have any more pressing stories on the go, I made a start.
Sixteen months and 14 front-page stories later, Google had lost around $800million (£586million) by one estimate, firefighting a scandal that saw hundreds of advertisers flee the company's video platform. Our stories exposed how YouTube placed hundreds of adverts for mainstream brands on videos showing horrific content, including terrorist and Nazi propaganda.
The ads were being placed on the videos without the brands’ knowledge, thanks to complex algorithms that matched online adverts to the person browsing rather than to the videos themselves. This meant that the first time Mercedes heard that ads for its new E-class were appearing on an Islamic State flag-waving video was when I contacted it for comment. Even worse, under YouTube's terms and conditions, the videos’ creators shared the profits with Google – meaning big name brands were inadvertently funding these extremists, who could earn significant sums from the most popular videos under the site's generous revenue-sharing model.
Later scoops published by The Times in 2017 revealed how YouTube and Facebook had failed to tackle child pornography on their platforms. In dozens of cases, the sites had refused to remove inappropriate content including apparent child prostitution. In some, the sites had actively promoted such content to users who had been identified by the software as “interested” in this material.
Let's go back to the end of 2016, however, when I was tramping the streets of Holborn, central London, interviewing mid-level media agency executives, trying to work out how digital advertising worked and I could possibly write a story about it. The first thing I learned was that the industry was huge. Last year, advertisers spent more on digital channels than on television – around $209billion worldwide. By 2020, digital ads are expected to make up 50 per cent of all ad spending. Google and Facebook eat up the vast majority of this pie, leaving other publishers the scraps.
Thousands of tech firms have sprung up to steer a digital advert between the brands wanting to advertise and the media owners ready to sell space. Many of them come under the umbrella of “ad tech”. Every time an advert is posted on a website, these middlemen take a cut. Such is the leakage that for every pound spent by a brand on digital advertising, only 30p filters across to the publisher.
Strange Bedfellows
Sitting amid pool tables and espresso machines, my agency sources explained how, despite its heft and complexity, the market for digital advertising is largely unregulated and prone to abuse. One source showed me an internal document from his agency's UK media-buying division. It showed that this particular agency was receiving huge rebates, equating to half of its overall income, from publishers to whom it directed spending.
In other words, the agency was getting paid to place ads in a particular place and not passing these payments back to the client. It wasn't disclosing such rebates to the brand either, the sources said, meaning that it had an incentive to spend the client's cash in a way that wasn't necessarily in their best interests.
Next, the source pointed to another category on the accounts called “other income”, which he described as “dark profits” made from the agency's programmatic unit. Programmatic advertising is the process by which complex, computerised tools track potential customers around the web, serving them relevant ads wherever they are browsing. If your computer has determined you are a city trader, based on your search history and a million other data points, it might offer up an ad for a Rolex even if you are looking at Lidl's website. To a programmatic advertiser, context is irrelevant and eyeballs are everything.
The leaked document showed me how profitable programmatic units were to this agency – bringing in about half what it was transparently making from fees. This was partly because, the source said, programmatic advertising is so complex that clients could be pushed into spending a huge amount of money on digital products without understanding its true value.
What I had now was evidence of potential sharp practice on behalf of a major advertising agency. The figures backed up some allegations made in a US industry report last year, which accused agencies of taking rebates and pushing customers into programmatic media to boost their profits. But where was the wider harm to consumers? This was all a little insider-baseball. I left the meeting feeling like I had cracked a big story but I still wasn't excited. Then, a few days later, I called the source back and asked him about harm. He said agencies were selling clients advertising on bad websites on the basis that the brands would never know.
A bit like in the credit crisis, when crappy mortgages were bundled together and traded, views on crappy websites could be bundled together with those on premium websites and sold to unsuspecting brands.
What sort of websites, I asked. Fake news websites, porn sites and other terrible media, he replied. Marketing directors only care that their brand can get in front of 20,000 18- to 35-year-olds, he said. If an agency offers to do this for cheap, the marketing director will be congratulated. There is no incentive for either agency or director to check where the ads are ending up.
I had now established that brands were effectively paying to advertise on websites they had never heard of – and probably would hate to be associated with. Once I had got my head around that, it was just a question of finding examples. We found hundreds. On fake news websites, on porn sites, piracy sites and, most controversially, on content posted on YouTube. By far the greatest number of ads we found next to inappropriate content appeared on Google's huge video platform.
Mercedes ads appeared on videos posted by IS supporters. Adverts for a Sandals resort appeared on a YouTube video put out by al-Shabab, the East African terror group. Ads for Argos appeared next to Combat 18 videos. This all went into the first splash in February 2018.
We published another front page the next day, showing how public institutions such as the National Theatre were advertising on fake news sites. At that stage, none of the brands pulled out. Most said they would look carefully at the allegations and “review” their relationship with media owners. Google issued a bland statement saying it had “zero tolerance” for violent content.
At that point, we checked if any government departments had posted ads placed next to terrible content. There were. An ad paid for the Home Office appeared on a YouTube video posted by Michael Savage, the controversial US shock jock. In 2009, he had been banned from the UK by the Home Office. Other extremists making money from both the British government and the UK's leading brands include David Duke, an anti-semitic holocaust denier and Steven Anderson, a pastor from Arizona who has called on all gay people to be killed.
As a result of our story, the British government pulled its advertising from Google and told us it was summoning the company's executives to explain themselves. From that point, things snowballed. Havas, a top-six ad agency that represents more than 240 UK brands, pulled its business from YouTube a day later; at the same time we published a front page showing that Google had failed to remove dozens of anti-semitic videos despite promising to review unacceptable content within 24 hours.
The Regulators Wake Up
Over that weekend in March, the boycott was joined by major multinationals such as Marks & Spencer, Tesco, Volkswagen and Toyota. Perhaps the big breakthrough came later that week, when US brands such as AT&T and Verizon, PepsiCo and Walmart pulled ads after we found they were placed against content including Animal Liberation Front videos.
This took the boycott global and prompted headlines around the world. One analyst put out a note suggesting that the scandal could cost the company $800million. That's a rounding error on Google's overall revenue. But it still mattered, because it represented a threat to Google's business model, which is based on attracting a constant stream of advertising in a regulatory-lite setting.
In fairness to YouTube, it took significant steps to clean out its stables. In January, it raised the threshold for monetisation of videos to 1,000 subscribers and 4,000 “watch hours” per month. All its preferred content, the premium content beloved by advertisers, would now be reviewed by human eye, it said. The zeal of its clean-up operation led to an uproar by thousands of “demonetised” users who labelled the move the “adpocalypse”. One YouTube user, incensed by demonetisation, sneaked into the Californian headquarters of the company in April and opened fire with a handgun, wounding three employees before killing herself.
Much of journalism is about timing. It was fortunate that The Times‘ revelations came amid a backlash against tech that is still gathering pace today. After years of relatively light regulation, the tech giants were facing a backlash – from the public, from businesses and from regulators such as Margrethe Vestager, European commissioner for competition – all of whom began to wake up to the downsides of automated, algorithm-based business models that trade entirely on the ability to hoover up data on users and sell that data to advertisers. This year, pressure has increased still further with the revelation that Cambridge Analytica harvested 87 million profiles from Facebook.
In the round, YouTube is going from strength to strength. In spite of the scandal, and in the face of a general backlash against tech companies, analysts predict the video site will generate more revenue this year than half the companies listed on the S&P 500 index.
In purely numerical terms, therefore, the damage caused by the “adpocalypse” revelations was no more than a pinprick in the side of an elephant. However, the real danger to these firms is that the torrent of negative stories will tip regulators and legislators over the edge. Regulation could fundamentally endanger these companies’ business models. They know this, which is one reason why Google alone spent $18million lobbying politicians in 2017. It is regulation, rather than an isolated scandal, that is the tech giants’ worst nightmare.