The cookie that refused to die, and why it stopped mattering

The third-party cookie was scheduled to die in 2022. Then 2023, then 2024. Then Google announced it would not deprecate it after all, and by 2025 the Privacy Sandbox as originally pitched was effectively shelved. The most anticipated funeral in ad tech never happened.

Here is the strange part: it stopped mattering, and the reason is instructive.

The deadline did the work the deprecation never did

Five years of deprecation threats forced the industry to build as if cookies were already gone. Server-side tagging became standard practice. Conversion APIs became table stakes on every platform. First-party data went from conference topic to line item. Consent management matured because EU regulators started writing checks nobody wanted to cash.

By the time Google blinked, the ecosystem it was protecting had already been replaced underneath it. The cookie survived, but the architecture that depended on it did not come back.

What the third-party cookie still cannot see

Even alive, the cookie was already blind where it counts. Safari and Firefox block it outright, which in our EU and US markets is a large share of high-value traffic. iOS strips identifiers at the OS level, and link tracking protection trims query parameters that survived everything else. Cross-device journeys, the normal way humans shop, were never in its field of view.

So the practical question in 2026 is not whether cookies exist. It is what share of your revenue you can attribute with infrastructure you own, on a domain you control, with consent you can prove.

The stack that won

For us, and for most operators we respect, the answer settled into three layers: a first-party pixel served from your own domain, server-side event delivery into the platforms through their conversion APIs, and an identity graph that ties sessions to people without depending on anyone’s browser policy.

That last layer is the moat. Browser vendors change policies annually. Platforms change APIs quarterly. A graph built on your own first-party events is the only part of the stack where the rules cannot be changed for you overnight.

We built ours because we had to, and it now carries 4B+ events a month for our own brands and for ScaleTrack customers. The cookie drama will produce more headlines. Own your data layer and they become entertainment instead of risk.