The original January 2024 outlook and what changed
Historical context: this article was published on 15 January 2024, when Chrome was testing restrictions on third-party cookies. Its original prediction of a general cookie phase-out did not become the plan described below. The status of the official sources was checked on 7 October 2026; the original publication date remains unchanged.
Table of Contents
In April 2025, Google retained Chrome’s existing third-party cookie choice in settings and said it would not introduce a new standalone cookie prompt. Later that year, it announced the retirement of several advertising APIs. For marketers, the useful question is how much of each conversion journey is actually observable, rather than when a single forecast deadline will arrive.
What are third-party cookies?
A cookie is text data stored by the browser, not executable code. A third-party cookie comes from a site different from the one the person is visiting. It can support cross-site functions such as advertising or an embedded service, but its availability depends on browser behavior and settings. See Google’s explanation of third-party cookies.
What happened to FLoC and FLEDGE?
The original article mixed earlier proposals with the then-current roadmap. These names now belong in the historical record:
- FLoC: Google’s January 2022 Topics announcement had already replaced the FLoC proposal, before this article was published. FLoC was not a new 2024 replacement technology.
- FLEDGE: Google renamed it Protected Audience API in April 2023. It was an on-device auction approach, not the same mechanism as grouping users into FLoC cohorts.
Why does cookie coverage still matter?
Cookie availability can vary by browser, user settings and enterprise policy. That means a campaign report can lose part of its observable journey without the same change in customer demand. Audit the functions and events your site relies on, test what happens when third-party cookies are blocked, and record the reporting coverage before comparing periods.
Chrome cookie choice and retired APIs
In its 17 October 2025 update, Google announced retirement of Topics, Protected Audience and Attribution Reporting, among other listed technologies. It also said it would continue supporting CHIPS and FedCM. These are specific API decisions; they do not mean every Privacy Sandbox technology was removed at once or that all cross-site cookies disappeared.
Do not build a new measurement plan around the old promise that FLoC or FLEDGE will universally replace cookies. Check the official status of the particular API or browser feature on which an implementation depends.
Measuring campaigns when signals are missing
Consider a visit to a sports-equipment shop followed by an ad on another site. Whether an advertising system can connect those visits depends on the signals available in that context. The historical illustration below is not evidence that every browser now uses a single cohort-based replacement.
For a useful performance readout, separate directly observed outcomes from modeled or attributed estimates. Compare the same dates, event definitions and conversion windows, and make coverage gaps visible. A missing connection between two visits is not proof that the advertising had no effect; an attributed conversion is not proof of an incremental sale.
Kevin Dosanjh, B2B Marketing Lead at Admiral Media, analyzes the situation and explains that finding a balance where users feel confident in their privacy while keeping the web ecosystem vibrant and accessible is crucial. His analysis underscores the challenge of maintaining an open web that respects user privacy. Dosanjh’s perspective highlights the complexity of adapting to new privacy standards without hindering the functionality and accessibility of the web.
Practical next steps for marketers
- Define the business outcome. Keep visits, accepted enquiries, qualified opportunities and revenue separate rather than treating a tracked click as a sale.
- Audit the visitor path. Check the important site functions and permitted events when cookie settings or consent choices vary, and document which reports contain modeled estimates.
- Test the decision that matters. Use an appropriate controlled comparison when you need to establish added value. Our incrementality guide explains why attribution alone cannot answer that question.
