Consent Mode Explained to a Child
Consent Mode v2 Explained: What Changes for Marketing Measurement in the EEA
How Reliable Are the Numbers You See Today in Google Ads and GA4, Now That Consent Mode v2 Is Mandatory for Those Operating in the European Economic Area?
Less than you would like. From the moment a user rejects cookie consent, Google stops observing the conversion and begins estimating it with an internal statistical model. With an average consent rate in Italy between 35% and 55%, a significant portion of the ROAS you read in your dashboard is not measured: it is inferred. Consent Mode v2 is not the problem itself. It is the signal of a structural change in how marketing will have to measure its effectiveness in the coming years.
What Consent Mode v2 Does, in Two Lines
As of March 2024, Google requires Consent Mode v2 for all advertisers operating in the EEA. When a user accepts cookies, tracking works as usual. When they reject them, the tag does not turn off: it continues to send anonymous pings without cookies. Google uses these pings to estimate missing conversions through proprietary statistical models.
The number you read on the platform is therefore a composition of two different ingredients: an observed part and a modeled part. The proportion depends on your site’s consent rate.
The Real Impact on the CMO's Work
For a Marketing Director, this changes three things simultaneously.
Reliability of numbers. If half of the conversions you see on Google Ads are produced by a model whose specifications you don’t know, the ROAS you bring to the Board contains an uncertainty that is difficult to quantify. You don’t know where the data ends and the estimation begins.
Comparability between channels. Google data is partially modeled with one logic. Meta data follows another logic, with its own CAPI and attribution models. TV and OOH data live in a world of their own. Putting these numbers together in a single effectiveness table is an exercise that seems useful but produces fragile conclusions. You are adding up measurements obtained with incompatible methods.
Budget planning. If the database on which you decide allocation is partly reconstructed statistically, every upstream model revision (i.e., every Google update) translates into shifts in your historical numbers. What was a performance in March may no longer be one in August, without anything having changed in the reality of the business.
The Three Possible Responses, in Order of Depth
A structural problem requires structured responses. For a CMO who wants to protect the quality of their measurement, operational options are arranged on three levels, from the most tactical to the most strategic.
Level one: raising the consent rate. A well-configured consent management platform, with clear copy and non-aggressive UX, can bring the consent rate from 35% to 55-60%. It is the most immediate way to reduce the share of modeled conversions. It remains a mitigating measure: it does not solve the problem, it reduces it.
Level two: strengthening first-party data. Building a direct relationship with the customer (login, loyalty program, structured CRM) reduces dependence on third-party cookies. It is a long-term investment that pays off beyond measurement: it also changes the quality of activations and personalization. However, it requires time, governance, and a data culture that many companies do not yet have.
Level three: measuring upstream of the consent problem. There are measurement methodologies that, by design, do not need to know if John Doe accepted cookies. They work on aggregated sales and media investment data: how much you spent on a channel in a week, how many sales you recorded in the same period. Period. Marketing Mix Modeling and lift tests fall into this family. They do not replace platforms; they complement them, and they do so with a logic independent of individual consent.
The three levels are not alternatives. They are complementary, and each responds to a different part of the problem. A CMO who decides not to move any of the three levers is choosing, consciously or not, to entrust the quality of their measurement to the proprietary statistical models of third parties.
The Arrow Every CMO Should Have in Their Quiver
No single tool is the answer to everything. A well-equipped CMO works with a composite measurement system, where each method covers the blind spots of the others. Platforms remain useful for day-to-day tactical optimization. First-party data feeds CRM and activations. Causal measurement, done with methods that live upstream of consent, provides the level of insight that the Board can take into strategic planning.
In this quiver of arrows, a well-constructed causal measurement (whether through Marketing Mix Modeling or targeted lift tests) is the one that remains valid even when the consent rate changes, even when a platform updates its model, even when a new regulation changes the rules of the game. It is the arrow with the longest range.
At Core Analytics, we have been building measurement systems that work at this level for over twenty years. Consent Mode v2 has not changed our approach: it has made it even more relevant for CMOs who need numbers that are stable over time, comparable across channels, and independent of variables they do not control.
The Real Point, Beyond Consent Mode
Consent Mode v2 is not an isolated development. It is a symptom of a direction that European marketing has already taken: user-level measurement is becoming progressively less reliable, and no technological update reverses this trend.
Those who build a measurement system today on foundations independent of individual tracking are not making a conservative choice. They are making a long-term choice, in line with the direction that the European regulatory framework will continue to chart.
The numbers you bring to the Board are only as valuable as they are stable over time. A measurement that depends entirely on signals that the regulator can modify at every iteration is not a solid basis for planning the next three years of budget.
It is worth stopping for a moment and asking yourself: is your marketing measurement system designed for the world that was, or for the world that is coming?
Leave a comment