Planning without the cookie countdown

There is a slide in most marketing roadmaps that has now been orphaned. It says something like "cookie deprecation readiness" and it sits in a quarter that was chosen because a browser had announced a date. Under it are four workstreams that were funded largely because of that date.

Google has confirmed that third-party cookies will stay in Chrome, and that the standalone choice prompt it proposed last July will not ship. The clock that half the industry pegged its plans to has simply stopped. Not paused, not moved again. Removed.

The temptation is to cancel the workstreams and move on. That would be a mistake, but the honest version of the argument against it is not "the cookie is dying anyway". It is that the deadline was never the reason. It was a convenient way to get funding for work that was justified on its own terms, and now those terms have to be stated out loud.

What the confirmation actually removes

It removes one thing: a forced migration date in one browser. That is worth acknowledging clearly rather than downplaying, because your paid media colleagues have real relief here and pretending otherwise damages your credibility with them.

It does not remove any of the following, all of which were true before the announcement and remain true today.

Safari and Firefox continue to restrict third-party cookies by default. Whatever share of your traffic those browsers represent is already unaddressable by that mechanism, and that share is not small in most consumer categories.

Mobile app tracking remains governed by platform permission prompts, which have nothing to do with browser cookies and are not affected by this news.

Consent requirements still apply. A cookie you are technically able to set is not a cookie you are permitted to read, and enforcement has been tightening independently of anything Chrome does. We went through the practical implications of that in consent signals a year on, and none of it changes because Chrome changed course.

Identity resolution across your own properties is still your problem to solve, because it always was. No browser was ever going to do that for you.

The four reasons that were never about cookies

If the work is going to continue, it needs reasons that would have held anyway. In our experience four of them do.

Your own data is the only input you fully control. Everything else is rented. A platform can change a permission model, a partner can lose a data source, a browser can change its mind in either direction. First-party data is the part of the stack that does not move underneath you.

Retargeting the same people repeatedly is expensive, not effective. The audiences that third-party cookies build most easily are existing customers and recent visitors, which is exactly the group you can reach through owned channels at near-zero marginal cost. The cookie question was always partly a cost question wearing a compliance costume.

Personalisation quality depends on depth, not breadth. Knowing that somebody browsed a competitor is weak. Knowing their size, their reorder interval and their service history is strong, and none of that arrives via a browser cookie.

Measurement was already unreliable. Multi-touch attribution across platforms has been degrading for years for reasons that predate this news. That problem needs a different method, not a rescued cookie.

Replace the deadline with a measurement commitment

A deadline is a poor motivator but an effective one. If you remove it, you have to put something in its place, and the thing that works is a recurring commitment to measure the programme properly.

Two methods carry most of the weight, and they answer different questions.

Marketing mix modelling works at the level of channels and spend over time. It is tolerant of missing user-level data, which is precisely why it is back in fashion, and it will not tell you anything useful about an individual customer journey. Treat it as a budget allocation tool, not a campaign optimisation tool.

Incrementality testing works at the level of a specific decision. Hold out a group, run the thing, measure the difference. It is slower and narrower than a dashboard, and it is the only method that answers "did this cause anything".

Commit to a calendar rather than a project. Two or three incrementality tests per quarter, each with a named owner and a decision attached. A model refresh on a fixed cadence. Put the dates in the plan where the deprecation milestone used to be, so the slot in the roadmap stays occupied.

Spend the reclaimed time on the data you already have

Most teams that were preparing for deprecation built an inventory of dependencies and then paused. That inventory is still the most valuable artefact you own, because it tells you where the programme is borrowing identity from somewhere it does not control.

Finish it, then work on three things.

Improve the quality of what you collect at the moments when people are willing to tell you something. Preference centres, onboarding questions, post-purchase surveys and sizing tools all produce zero-party data, and the trade is honest as long as the answer visibly changes what the customer receives. It usually does not, which is why response rates fall.

Fix identity across your own properties: web, app, till, call centre and marketplace. This is unglamorous joining work and it produces more personalisation capability than any vendor purchase.

Then reduce what you are carrying. The audience libraries that accumulated during deprecation planning are typically full of overlapping definitions nobody uses. The method for clearing them out is the same one we described in pruning segments, and doing it now is cheaper than doing it after the next platform migration.

What to tell the board

Be straightforward. The migration deadline has gone, the work continues, and here is why in one sentence: we are reducing the share of our marketing that depends on permissions other companies grant us.

Then show the number that makes it concrete. Take revenue attributable to audiences you can reach through owned channels, divide by total marketing-driven revenue, and track it quarterly. Define it once, write the definition down, and do not change it when it moves in an unhelpful direction.

The uncomfortable part is that without a deadline, this programme now competes for budget on merit against things with faster returns. That is a fair fight and it is winnable, but only by teams that can produce evidence rather than a countdown. Start assembling the evidence this quarter, because the next time the argument comes up, "it was on the roadmap" will not be enough.