The advertising cartel coming to your web browser
01 Jun 2026
The advertising cartel coming to your web browser
When Meta, Google and Apple agree on a “privacy”<br>feature, watch out.
The three companies (along with Mozilla, which is on one of their “ad<br>features in the browser” kicks again) are drawing up a built-in<br>advertising measurement system, called Attribution Level 1, as a<br>standard feature of web browsers. The system is intended to measure the<br>effectiveness of advertising by enabling advertisers to correlate<br>“impressions,” the occasions on which someone saw an ad, with<br>“conversions,” when people bought something.
Don’t look for a section on permissions or consent in that document,<br>by the way. There isn’t one. And nothing about nerd lawyer stuff like<br>“opt out of sale” or “objections to processing” in there, either. The<br>Big Tech companies want a two-track system, where other<br>companies’ ad features are required to do all the privacy regulation<br>hassles, but the browser’s own built-in tracking feature is<br>something that people have to find<br>the right setting for and turn off.
Unfortunately, this is not just a chapter in Big Tech’s ongoing<br>antitrust saga. The attribution cartel is on track to perpetrate real<br>harms to users, including:
Built-in advantage for search, social, and app store<br>advertising: More money for Big Tech, less for legit sites and<br>other ad-supported resources.
Added incentives for riskier tracking:<br>Obfuscating the source of a sale makes it easier to get a payoff from<br>tracking practices that would be seen as problematic on their<br>own.
Those consequences are unavoidable because of the proposal’s narrow,<br>mathematical privacy goals, which are a mismatch for the kinds of<br>privacy harms that people experience in the real world. In the “Privacy<br>Considerations” section, the proposal says,
The main privacy goal of this API is to ensure that providing sites<br>with the ability to perform attribution does not improve their ability<br>to perform cross-site recognition.
The system is supposed to produce aggregated measurements while<br>making it prohibitively difficult for an advertiser to discover whether<br>any one person who bought something is the same person who saw an ad.<br>Technically, the way it works is that a script running on a site with<br>ads asks the browser to record an ad impression. Then the browser keeps<br>a record of ads seen from all the sites you visit. Later, when you buy<br>something, the retail site can ask the browser to generate a “conversion<br>report” that can be passed to a centralized aggregation service. The<br>aggregation service can then give the site some aggregated results, in a<br>way that does not reveal whether any individual who bought something<br>ever saw a particular ad or visited a particular site.
So why are the same companies that are notorious for tracking people<br>so fired up about it? The problem is that the attribution tracking won’t<br>be functioning in isolation. It has to interact with other technologies<br>and business models. Even if the browser developers can pull off their<br>ambitious goal of preventing “cross-site recognition,” the proposal<br>would make life worse on the real Internet.
Problem one: Over-rating search, social, and app store ads
Marketing data expert Rick Bruner explained<br>this best.
Lower-funnel media naturally appear more effective because they<br>intercept demand after it has already been created elsewhere. Search is<br>the classic example. Brand advertising may create the demand, but the<br>search click gets the attribution credit because it occurs closest to<br>the sale.
And the shift isn’t just a matter of a Big<br>Tech squeeze on smaller companies, or an oligopoly tax on everything<br>you buy. The attribution cartel threatens us as citizens, too. Many<br>advertising-supported media have positive externalities—they benefit<br>even people who don’t use them. Corruption<br>thrives where newspapers shut down. Meanwhile, the negative<br>externalities of Big Tech are too numerous or too content-warning-worthy<br>to list here. The average person in the USA has about $1200/year spent<br>on advertising intended to reach them. Where do you want “your” $1200<br>spent?
Problem two: Incentives for extra tracking
The attribution cartel makes it impossible to tell an ad that<br>actually sold something apart from an ad that got placed in front of<br>someone right when they were about to buy.
And actually selling something is hard, so it’s much easier to “win”<br>attribution by identifying likely buyers and targeting ads to them. The<br>Big Tech companies can do this with machine learning, at a scale that<br>makes the real effects hard to sort out. A company with access to, say,<br>an always-on mic on a “smart TV,” would have trouble selling keywords<br>from people’s conversations directly. But the attribution cartel makes<br>it...