Online advertising was sold for two decades on a promise of .
An advertiser would reach exactly the person likely to buy, and waste would disappear.
Delivering that promise required data about individuals collected across unrelated websites.
made the market possible and made it impossible to explain to a user.
Regulation and changes have now removed much of that capability.
The measured consequence for publishers has been smaller than the industry predicted.
advertising, which matches an advertisement to the page rather than to the person, performs close to the tracked version in several controlled comparisons.
It also requires no personal data, which removes an entire category of legal risk.
Why the industry did not move there earlier is a question about incentives rather than technology.
An paid for matching individuals has no reason to prove that matching pages works equally well.
Measurement is the deeper problem and it survives every change of method.
A large share of reported advertising effect is rather than persuasion.
Showing an advertisement to people already searching for the product produces impressive figures and proves very little.
Controlled experiments run by large advertisers have repeatedly found far below the platform's own estimate.
The industry's accounting problem is therefore older than its privacy problem and has attracted a fraction of the attention.