The targeting methods that defined digital advertising for a decade depended on identifiers that followed people between sites. Regulation and platform policy removed most of that capability, and the industry restructured around the gap.
Consent turned collection into a choice
Privacy regimes in several major markets require an affirmative choice before non-essential tracking may occur, presented in a way that permits refusal.
A substantial share of visitors decline, which means the data available for targeting and measurement now covers only part of the audience.
The proportion varies by market, by device and by how the request is presented, which introduces bias as well as loss into every dataset built this way.
Browsers acted independently of regulators
Several browsers restricted third-party cookies and limited the lifespan of other storage mechanisms as a default rather than as a user choice.
Mobile platforms introduced permission prompts for cross-application tracking, and the majority of users did not grant it.
These changes affected far more traffic than regulation alone, because they applied to everyone rather than to a jurisdiction.
Contextual targeting returned with better tools
Placing advertisements according to what a page is about never depended on identifying the reader, so it was unaffected by any of these changes.
The method had been considered crude, but language processing improved its precision considerably, allowing distinctions between topics that keyword matching confused.
It also handles the case that behavioural targeting handles worst, which is reaching someone whose interest exists right now rather than in their history.
First-party data became a competitive asset
Relationships where a customer identifies themselves, through accounts, subscriptions or purchases, produce data that remains usable under the new arrangements.
This advantaged businesses with direct customer relationships and disadvantaged those reaching customers only through intermediaries.
It also shifted investment towards loyalty programmes and registration mechanics, which are now partly justified as advertising infrastructure.
Measurement moved towards modelling
With individual paths no longer observable end to end, platforms began estimating the missing portion and reporting the estimate alongside observed events.
Aggregate methods that were previously considered old-fashioned, including holdout testing and marketing mix analysis, regained relevance because they never required individual tracking.
The practical result is measurement that is less precise about individuals and, in several respects, more honest about what advertising actually caused.
That trade has been uncomfortable for teams whose reporting was built on granular attribution, since the replacement produces ranges and confidence intervals where a single confident number used to sit.