Consent Mode v2 Is Fully Enforced: What EU Brands Lost and How to Measure Anyway
Google now fully enforces Consent Mode v2 across the EEA, and most EU stores directly measure a minority of their traffic. Here is what each measurement layer still sees, and why a causal read on consented data beats a modeled guess.
Three platforms, three answers, one budget to defend. Your GA4 export already holds the answer, and the first finding is free and never leaves your browser. Get your first finding free
By Joris van Huët, Founder & CEOUpdated 7 min read
Run the numbers for your store: the free server-side tracking ROI calculator.
Since June 2026, Google fully enforces Consent Mode v2 across the European Economic Area: no valid consent signal means no Google tag activity, no remarketing audiences, and no ad personalization measurement for that user. GA4 covers the hole with behavioral and conversion modeling, which is an estimate dressed as a measurement. The workable path is narrower and more honest: measure what your consented first-party data actually proves, and use causal inference for budget decisions.
What did Consent Mode v2 enforcement actually change in June 2026?
See also: How to Set Up Consent Mode v2 on Shopify
Consent Mode has existed for years as the signaling layer between your consent banner and Google tags. Version 2 made two signals central: ad_user_data, which governs whether advertising data may be collected, and ad_personalization, which governs remarketing and personalized ads. Google had pushed EEA advertisers toward adoption since 2024, but June 2026 is the hard stop. Full enforcement means tags that arrive without a valid consent state lose advertising functionality entirely: no audience building, no personalization, and degraded conversion measurement for that user.
For marketers, the useful framing is that this is not a reporting tweak like an attribution window change. It is a supply cut. The raw material every downstream tool consumes, tag-level event data from real browsers, now exists only for visitors who actively said yes. Everything built on top of it, audiences, bidding signals, attribution, inherits that constraint.
It also did not arrive alone. June 2026 landed in the middle of a year full of platform measurement shifts, from Meta's reporting overhaul to Shopify's plumbing changes. We track those separately in our 2026 attribution changelog; this post stays focused on the EU consent problem and the playbook that still works.
What enforcement does not take away: your own first-party data. Purchase records, email engagement, on-site behavior from consented sessions, and your GA4 export remain yours. The loss is specific to Google tag-dependent measurement of non-consenting users, which is why the fix starts from what you still hold rather than from what Google stopped sending.
How much data did EU brands actually lose?
Do the arithmetic on an illustrative store with 300,000 monthly sessions and a 55 percent decline rate: 135,000 consented sessions now carry all of your measurement. Every pixel audience, every GA4 segment, and every optimization signal your ad platforms receive is built on that remaining 45 percent.
GA4's answer is modeling. Behavioral modeling estimates what declined users probably did, trained on the behavior of similar consented users. Conversion modeling estimates conversions that were never directly observed. Both keep dashboards looking reassuringly full. Neither is an observation, and GA4 does not show you the modeled individuals or the exact size of the estimate behind each number.
There is also a selection problem that no model removes: people who decline tracking are not a random sample of your visitors. They tend to be more privacy-aware, and their purchase behavior differs from people who accept. An estimate trained on consented users and projected onto declined users carries that bias into every report, silently.
You feel the loss in specific places first. Remarketing pools shrink, because only consented users can join them. Automated bidding receives thinner signal, so strategies take longer to stabilize after changes. And new versus returning splits skew, because declined sessions look permanently new. None of this arrives as an error message. It arrives as numbers that quietly stopped meaning what they meant in May.
What does each measurement layer still see after enforcement?
Four measurement layers remain available to an EU brand, and they do not see the same reality. The table sums up the post-enforcement position for an illustrative EU store spending €150K per month.
| Measurement layer | What it still sees after June 2026 | How it fills the gap | Directly observed share (illustrative, July 2026) |
|---|---|---|---|
| Browser pixel (client-side) | Consented sessions only | Nothing; declined traffic is simply invisible | Under 50 percent |
| GA4 with Consent Mode v2 | Consented events plus modeled sessions and conversions | Behavioral and conversion modeling | Under 50 percent observed; the remainder is estimated |
| Server-side tagging | Cleaner consented first-party events, less loss to browser restrictions | Signal quality, not extra consented volume | Same consented share, measured more reliably |
| Causal read on first-party data | What consented data proves about which spend caused revenue | Causal inference over observed outcomes, with uncertainty stated | Read from observed data only |
Two clarifications matter. First, server-side tagging is genuinely worth doing, but understand what it buys: better durability and quality for consented data, not a larger consented universe. Second, treat any vendor selling consent-proof tracking as a compliance incident waiting to happen. Recovering data that visitors refused to share is a GDPR violation with a pricing page, not a measurement strategy.
Why is a modeled guess not a measurement?
Because two opacity layers now stack on top of each other. Then data driven attribution distributes conversion credit across the touchpoints that remain, using a model whose inner workings Google does not publish. The output is a precise-looking number assembled from two estimation steps, and you cannot audit either step. We took the second layer apart in our review of GA4's attribution black box.
To be fair, modeling has a legitimate job: keeping automated bidding fed with enough signal to function. But a number built to steer Google's algorithms is not automatically a number you should hand your CFO. When modeled traffic and modeled credit disagree with your bank account, there is no panel you can open to find out why. You are asked to trust the machine that profits from your trust.
The discipline that survives enforcement is different in kind: take only data you actually observed, state your assumptions out loud, and put an uncertainty range around every conclusion. Less comfortable than a full-looking dashboard. Far harder to be wrong in a way nobody catches.
How do you measure anyway? The consented-data playbook
1. Earn consent honestly. Banner design, timing, and a clear value exchange move consent rates more than any tracking trick. On the illustrative 300,000-session store, every extra point of consent rate is 3,000 additional measured sessions per month. That is the cheapest measurement upgrade available to you, and it compounds. Stay inside GDPR lines: dark patterns are regulatory debt with interest.
2. Move tagging server-side. You keep the same consented universe, but you lose less of it to browser restrictions and ad blockers, and you control exactly what leaves your infrastructure and goes to each platform.
3. Label modeled numbers as modeled. Separate observed from estimated data in your reporting wherever the tools allow it, and never present a modeled ROAS figure to stakeholders as a measured one. The fastest way to lose a CFO's trust is a number you cannot defend.
4. Make budget decisions on causal reads, not modeled credit. Where you can run them, incrementality tests answer the causal question directly. Keep your attribution window consistent when comparing periods, or you will mistake ruler changes for performance changes.
This last step is what Causality Engine was built for. You export your GA4 data, and we return a causal read in 1 to 2 minutes: which channels caused revenue, which merely showed up near it, and how confident the read is. €99 per read, €299 per month for Pro, no pixel to install, no annual lock-in. Run a causal read on your own GA4 export.
Key takeaways
- Consent Mode v2 has been fully enforced across the EEA since June 2026; without valid consent signals, Google tags lose advertising functionality entirely.
- GA4 behavioral and conversion modeling fills the gap with estimates; stacked with data driven attribution, you get two black boxes, not one measurement.
- Server-side tagging improves signal quality for consented users but cannot recover declined ones; anything promising consent-proof tracking is a GDPR risk.
- Budget decisions belong on causal reads of consented first-party data, with uncertainty stated rather than hidden inside a model.
Further reading
Frequently asked questions
What is Google Consent Mode v2?
Consent Mode v2 is Google's framework for passing a visitor's consent choice from your cookie banner to Google tags. The v2 update put two signals at the center: ad_user_data, which controls advertising data collection, and ad_personalization, which controls remarketing and personalized ads. Since June 2026 enforcement is full in the EEA, so without a valid consent signal there is no advertising functionality for that user.When did Consent Mode v2 become mandatory in Europe?
Google moved Consent Mode v2 to full EEA enforcement in June 2026. From that point, tags arriving without a valid consent state lose access to advertising features, audiences, and personalization measurement. Earlier phases were softer pushes toward adoption; June 2026 is the hard stop that turned consent signaling from best practice into a requirement for measurement in Europe.Is GA4 conversion modeling accurate?
It is an estimate, and it should be treated as one. Behavioral modeling infers what declined users probably did based on similar consented users; conversion modeling infers unobserved conversions. The output is not auditable user by user, and Google does not publish the inner workings. It is useful for keeping automated bidding fed, but it is not decision-grade measurement.Does server-side tagging fix consent data loss?
No. Server-side tagging improves the quality and durability of the events you are allowed to collect, and it reduces losses to browser restrictions and ad blockers. It does not and should not recover data from visitors who declined consent. Any setup that promises consent-proof tracking is a GDPR compliance risk, not a measurement strategy.How can I measure marketing ROI when half my traffic is unmeasurable?
Work with what you can observe. Keep consent rates healthy with an honest banner, collect consented events server-side, and make budget decisions from causal reads on first-party data rather than modeled platform credit. Incrementality testing and causal inference on analytics exports answer the question that matters: which spend caused revenue, with uncertainty stated instead of hidden.
Go deeper: Incrementality testing, explained.
Sixty-second versions of these ideas: Causality Engine on YouTube Shorts.
Have an audience of ecommerce operators? Earn 20% recurring commission by recommending the €99 read.
Keep reading
Terms in this article
- AttributionAttribution identifies user actions that contribute to a desired outcome and assigns value to each. It reveals which marketing touchpoints drive conversions.
- Attribution WindowAttribution Window is the defined period after a user interacts with a marketing touchpoint, during which a conversion can be credited to that ad. It sets the timeframe for assigning conversion credit.
- Causal InferenceCausal inference is the set of methods for estimating how much an outcome changes because of a cause, rather than alongside it. It works by comparing what happened with an estimate of what would have happened without the cause.
- Dark PatternsDark patterns are user interfaces designed to trick users into unintended actions. These deceptive practices should be avoided.
- Data Driven AttributionData-Driven Attribution uses machine learning to analyze customer touchpoints and assign conversion credit. It determines the true impact of each marketing channel.
- ExperimentsExperiments are scientific procedures that test hypotheses or demonstrate facts. In marketing, experiments like A/B tests determine the causal effect of campaign changes, enabling data-driven decisions.
- IncrementalityIncrementality is the part of an outcome, such as orders or revenue, that a marketing activity caused: the extra results compared with what would have happened without it. It is a causal effect, not the number of conversions credited to a channel.
- PersonalizationPersonalization tailors a service or product to specific individuals or groups. In marketing, personalization increases conversions by showing relevant content and offers.