Google and Meta made a secret deal to target advertisements for Instagram to teenagers on YouTube, skirting the search company’s own rules for how minors are treated online.
According to documents seen by the Financial Times and people familiar with the matter, Google worked on a marketing project for Meta that was designed to target 13- to 17-year-old YouTube users with adverts that promoted its rival’s photo and video app.
The Instagram campaign deliberately targeted a group of users labelled as “unknown” in its advertising system, which Google knew skewed towards under-18s, these people said. Meanwhile, documents seen by the FT suggest steps were taken to ensure the true intent of the campaign was disguised.
The project disregarded Google’s rules that prohibit personalising and targeting ads to under-18s, including serving ads based on demographics. It also has policies against the circumvention of its own guidelines, or “proxy targeting”.
Meta’s YouTube campaign to pull in younger users to Instagram was already in development when Mark Zuckerberg made a dramatic appearance before US Congress in January, where the Facebook co-founder apologised to the families of children who had been victims of sexual exploitation and abuse on his platforms.
The Silicon Valley-based pair, who are normally fierce competitors as the world’s two largest online advertising platforms, embarked on the effort late last year as Google sought to bolster its advertising earnings and as Meta scrambled to retain the attention of younger users against fast-growing rivals such as TikTok. Last week, Zuckerberg told investors that a recent push to engage more 18- to 29-year-olds had been bearing fruit.
The companies worked with Spark Foundry, a US subsidiary of French advertising giant Publicis, to launch the pilot marketing programme in Canada between February and April this year, according to the people and documents seen by the Financial Times.
Due to its perceived success, it was then trialled in the US in May. The companies had planned to expand it further, to international markets and to promote other Meta apps such as Facebook, people familiar with the matter said.
While the pilot programmes were small, Google saw them as an opportunity to grow into a more lucrative “full-funnel” relationship with Meta that would involve more splashy and expensive “brand” adverts on YouTube as well as its other platforms.
When contacted by the FT, Google initiated an investigation into the allegations. The project has now been cancelled, a person familiar with the decision said.
Google said: “We prohibit ads being personalised to people under-18, period. These policies go well beyond what is required and are supported by technical safeguards. We’ve confirmed that these safeguards worked properly here” because no registered YouTube users known to be under 18 were directly targeted by the company.
However, Google did not deny using the “unknown” loophole, adding: “We’ll also be taking additional action to reinforce with sales representatives that they must not help advertisers or agencies run campaigns attempting to work around our policies.”
Meta said it disagreed that selecting the “unknown” audience constituted personalisation or a circumvention of any rules, adding that it adhered to its own policies as well as those of its peers when advertising its services. It did not respond to questions about whether staff were aware that the “unknown” group skewed to younger users.
“We’ve been open about marketing our apps to young people as a place for them to connect with friends, find community, and discover their interests,” Meta said.
Spark Foundry did not respond to multiple requests for comment.
Last week, the US Senate overwhelmingly passed a bill, the Kids Online Safety Act, that would place a duty of care on social media platforms to protect children from harmful online content, in a rare moment of bipartisan agreement that brings the US closer to major legislation targeting Silicon Valley on child safety.
“Big Tech companies cannot be trusted to protect our kids,” Republican senator Marsha Blackburn told the FT when contacted about the Google-Meta tie-up. She urged Congress to pass the Kosa bill. “They once again have been caught exploiting our children and these Silicon Valley executives have proven that they will always prioritise profit over our children.”
Jeff Chester, executive director of the Center for Digital Democracy, which advocates for child privacy, said: “Meta is bleeding young people and they’ve figured out a backdoor.”
Meta has long faced scrutiny for its policies on minors. It is being sued by 33 states accusing it of deploying ‘manipulative’ practices towards young users, which it denies. Meanwhile, the Federal Trade Commission is also seeking to ban Meta from making money from teen audiences as part of an update to an existing privacy settlement, which the company is challenging in court.
In 2021, it shelved plans to launch a kids version of Instagram following a public backlash and after whistleblower Frances Haugen leaked the Facebook parent’s own research suggesting the app is detrimental to the mental health of teenage girls.
How an unlikely partnership developed
According to documents and several people familiar with the matter, the Meta-Google project originated in early 2023 when Spark Foundry, acting for the Instagram parent, asked a range of partners to pitch for a “Meta IG Connects” advertising campaign.
Spark was working on behalf of the Meta marketing data science team and was tasked with getting more “Gen Z” customers to download Instagram, which has been losing users to rival apps, in particular TikTok, internal documents show.
Instagram has been fretting about losing its “teen foothold” for years. It previously allocated its entire marketing budget to targeting teenagers, in particular the 13- to 15-year-old “early high school” segment, according to a 2021 report by the New York Times.
In one email, seen by the FT, an ads manager at Spark asks Google to pitch for the campaign, specifically identifying the “primary” demographic to be targeted is “13 to 17” year-olds and requiring it to be measured by data collected directly from viewers. A secondary objective was 18- to 24-year-olds.
In 2021, Google introduced what it said were tougher protections for teenagers on its sites. “We will block ad targeting based on the age, gender, or interests of people under 18,” it said.
Google’s “ad-serving protections for teens” policy adds: “We expect all our advertisers to follow local legal requirements when using our products . . . as well as all Google Ads policies.”
But Google staff proposed a workaround to bypass the policy: a group called “unknown”, people familiar with the matter said.
On its website, Google says the “unknown” group “refers to people whose age, gender, parental status or household income we haven’t identified”. But staff at the internet group had thousands of data points on everything from users’ location via phone masts to their app downloads and activity online. This allowed them to determine with a high degree of confidence that those in the “unknown” group included many younger users, in particular under-18s.
Turning off other age groups for which they had demographic data left only the unknown group, with its high proportion of minors and children: it was described as a way of “hacking” the audience safeguards in their system, one of the people said.
“Targeting the ‘unknown’ category reaches a varied and wide audience of people,” including those who have ad personalisation turned off, Google said in response to questions about the use of the tactic to circumvent its policy.
Meta said: “Google’s ‘unknown’ targeting option is available to all advertisers — not just Meta — and we have clear principles we adhere to when it comes to how we market our apps to teens on other platforms.”
During the pitching process, another email from Spark in late 2023 asked Google to provide Meta with “platform-specific data and insights into teen behaviour”. This would “enable us to tailor and refine our media tactics, messaging and creative execution”, it read.
As part of its pitch, Google also boasted of its “really impressive” usage by 13- to 17-year-olds, handily outstripping daily engagement on TikTok and Instagram, documents show.
Google won the mandate from Spark and the teams on both sides took precautions, banning any direct reference to the age range in writing, one of the people said. Staff used euphemisms in presentations, such as slides with only the words “embrace the unknown”, according to documents reviewed by the FT.
Chester of the Center for Digital Democracy said of the tie-up between Meta and Alphabet-owned Google: “It shows you how both companies remain untrustworthy, duplicitous, powerful platforms that require stringent regulation and oversight.”
