In short
- Consent Mode v2 has been mandatory since March 2024 for any European advertiser that wants to keep remarketing and conversion measurement inside Google. It is not an optional refinement.
- Advanced mode only pays off above a volume threshold. Google Ads models conversions from 700 ad clicks over seven days, for a single domain and country pairing. Below that, advanced mode recovers nothing that basic mode would not.
- Many smaller companies never reach that threshold, and carry the complexity of advanced mode without ever collecting the benefit.
- The real lever sits elsewhere. One point gained on your acceptance rate is worth more than a badly wired advanced mode.
- And the subject goes beyond compliance. Without consent properly collected and transmitted, you are not deciding on less data: you are deciding on a sample whose distortion you cannot see.
The question almost always arrives in the same words: we are being told to switch to advanced mode, should we? The honest answer starts with another question that nobody asks first. How many ad clicks do you get in a week?
Because advanced mode is not a setting that improves measurement in the abstract. It is a mechanism that feeds a modelling engine, and that engine only switches on above thresholds Google publishes. Below them, you get the complexity without the benefit.
What Consent Mode v2 actually does
The principle fits in a sentence. Instead of blocking or loading Google tags, the page tells them the state of consent, and they adapt their behaviour accordingly.
Since March 2024, passing that state has been mandatory for any European advertiser using Google’s audience and measurement features. Without it, remarketing stops and imported conversions become unusable. So this is not a compliance question in the strict sense, compliance being the job of your banner: it is a condition of access to advertising features.
Two signals govern measurement: consent for analytics storage and consent for advertising storage. Two more, added by version 2, cover ad personalisation and data sharing with Google. That addition is what gave the version its name, and what made the update mandatory.
Basic or advanced: what each one really does
| Basic mode | Advanced mode | |
|---|---|---|
| Before the visitor answers the banner | No Google tag loads at all | Tags load, with no cookie and no identifier |
| If the visitor declines | Nothing is sent, the visit is invisible | A cookieless signal is sent, the visit is counted without being tied to anyone |
| If the visitor never answers and leaves | Nothing is sent | A cookieless signal is sent |
| What it enables | Generic modelling only, with no calibration on your own data | Model calibrated on your own account, above the thresholds only |
| What it costs | Simple to deploy, little room for wiring mistakes | More complex, and bad wiring sends data it should not |
The thresholds, and why they decide for you
Here is the point that vendor articles leave out, and that should open the conversation.
Conversion modelling in Google Ads switches on from 700 ad clicks over seven days, for a single domain and country pairing. On the Google Analytics 4 side, behavioural modelling requires roughly 1,000 daily events with consent declined for at least seven days, and 1,000 daily users with consent granted on at least seven of the last twenty-eight days.
Translate that into a real business. Seven hundred ad clicks a week in a single country implies an advertising budget that is both substantial and concentrated. A company spending across three countries spreads its clicks, and each domain and country pairing is assessed separately: total volume can look comfortable while no single market clears the bar.
The mistake that costs the most Switching to advanced mode, seeing no extra figures come through, and concluding the wiring is broken. In most of the cases we audit the wiring is fine: the volume is simply too low to trigger modelling. Weeks get burnt debugging an absence that is normal behaviour.
How to tell which side you are on
The calculation takes ten minutes, with no tooling and no developer. If you already have both figures in front of you, our modelling threshold checker gives you the verdict straight away.
- Open Google Ads and pull the last seven days of clicks, country by country. Not the total: the country breakdown. The threshold applies per domain and country pairing.
- Check whether any single country clears 700 clicks over the period. If none does, Google Ads modelling will not switch on, whatever mode you run.
- In GA4, compare your daily active users against the order of magnitude of 1,000. If you are in the low hundreds, behavioural modelling will not switch on either.
- Write down your acceptance rate. That is the figure that decides what happens next, and almost nobody tracks it.
If you sit above the thresholds, advanced mode earns its complexity and deserves careful wiring. If you sit below, it will give you nothing that basic mode does not, and your energy is better spent elsewhere.
What to do when you are below the thresholds
This is where most of the companies we work with sit, and it is not bad news. It simply means the lever is not a technical one.
The lever is your acceptance rate. Every point you gain on the banner brings whole visitors into your measurement, with their journey and their conversion, where modelling only ever produces an aggregated estimate. And that point is won through writing, not code: a purpose written in one comprehensible sentence is accepted more often than a list of technologies. We set out the method in our article on designing and writing a cookie banner.
The second lever is the reliability of what actually leaves the page. Consent can be properly collected and badly transmitted, with nothing showing up on any dashboard, and that kind of failure runs for months. We describe how to check it in our article on tracking that lies in silence.
What to take away Below the thresholds, modelling fills nothing. A visitor who accepts is worth vastly more than a modelled one, because they are measured for real, with their journey and their value. Working on the banner before the consent mode is the best ratio of effort to result, and it is also the only lever that belongs entirely to your company.
Beyond the legal obligation: consent decides what your figures are worth
Consent is almost always presented as a legal constraint, a box to tick to avoid an inspection. That misses its most useful day-to-day function: it is what determines whether your data describes your business, or an unidentified part of your business.
The reason is simple, and it still surprises a lot of boards. Refusals do not fall at random. They cluster on certain browsers, certain devices and certain traffic sources. A visitor arriving from your newsletter already knows you; a visitor arriving from a social ad discovers your brand at the moment the banner appears. Those two people do not answer the same way.
The consequence is that a sixty per cent rate does not cost you forty per cent of volume. It leaves you a sample whose composition no longer resembles your audience, and your channels are not measured with equal intensity. Two strictly identical campaigns can show very different acquisition costs for that reason alone, and budget migrates towards the one that is simply better measured. We set out the arithmetic, with a worked example, in our article on writing a banner people actually understand.
How to frame this internally A banner is not a legal device bolted onto a shop. It is the first link in your measurement chain. A company that treats its banner as an administrative formality is agreeing, without realising it, to steer budgets on a biased sample. That is why we work on banner wording inside measurement engagements, not only inside compliance ones.
The corollary is worth stating plainly, because it gets sold the wrong way round: a setup that “recovers” visitors who declined is not a measurement optimisation, it is a breach. Valid data is not obtained by working around the visitor’s choice, it is obtained by making sure they understand what is being asked of them.
The options, and what they are worth
| Option | What you get | Who it suits |
|---|---|---|
| Basic mode, carefully written banner | Compliant, simple, no risk of sending anything unintended. You measure fewer people, but what you measure is real | Anyone below the modelling thresholds, which is most smaller companies |
| Advanced mode, properly wired | Feeds modelling and recovers part of the lost conversions, provided you clear the thresholds | Advertising volume concentrated on one or two countries |
| Advanced mode, badly wired | The worst case: all the complexity, none of the benefit, and a real risk of sending data before consent | Nobody. It is nonetheless what we find most often in audits |
| No Consent Mode at all | Loss of remarketing and of imported conversion measurement inside Google since March 2024 | No case at all |
Tooling: the consent platform does half the work
Google does not require a certified platform from advertisers: you may keep your own banner, provided Consent Mode is implemented correctly. So the mechanics matter as much as the tool. The default state must be declared before any command that sends measurement data, otherwise tags collect before they know the visitor’s choice.
For our clients we deploy the modules published by Datafirefly Limited, our agency’s sister company. The same work exists on the three platforms we meet most often, which matters when a company runs several shops on different technologies.
| Your platform | The module | What it adds in its own right |
|---|---|---|
| WordPress and WooCommerce | DataFirefly Cookie Consent | Native Consent Mode v2, compatible with high-performance order storage |
| PrestaShop 8 and 9 | DataFirefly Cookie Manager | Built on the open source tarteaucitron.js engine, consent audit log, multi-store |
| Shopware 6 | DataFirefly Cookie Consent | Native Shopware extension, proof of consent retained |
What they have in common is the part that matters here: they emit all seven signals before the tag manager loads, and above all they audit the trackers actually loading on your pages. That is the only way to verify that third-party content, an embedded video or a chat script is not dropping anything before the visitor has chosen.
A word on the commercial model, because over three years it weighs more than people expect. The best known consent platforms run on a subscription indexed to volume: the more traffic you get, the higher the bill, and the compliance budget becomes a variable cost nobody planned for. The published grid for these platforms steps up by pageview tier, and the top tier moves to a custom quote. On the accounts we look after, the monthly line has run past a hundred euros. The modules listed above sit in the other model, the one-off purchase: you pay for the licence once, with twelve months of updates included, and your traffic can double without the budget line moving.
This is not an absolute argument. A subscription platform brings a pre-configured catalogue of services and pooled regulatory monitoring, which has real value for a company with no technical resource. But the question deserves to be asked before signing, and it rarely is: over the lifetime of your site, what does this banner cost, and does that cost track your growth without adding anything in return?
Method beats tooling: no module reports your acceptance rate for you, and none clears Google’s thresholds on your behalf. The module applies a decision cleanly, it does not make it.
What this says more broadly
Consent Mode illustrates a common drift in digital marketing: a mechanism gets adopted because it exists and because someone said to, without anyone checking whether it produces anything in this particular case.
The question to ask of any new piece of technique is the same. Above what volume does this work, and am I above it? It disqualifies a lot of recommendations, and it saves weeks.
Sources
The verifiable claims in this article link to their primary source, opened on 10 August 2026. We do not cite a source we have not read.
- Google Ads Help, About consent mode modeling. Consulter
- Google Analytics Help, Behavioral modeling for consent mode. Consulter
- Google for Developers, Consent mode overview. Consulter
- Google for Developers, Set up consent mode on websites. Consulter
- Google Ads Help, Updates to consent mode for traffic in European Economic Area (EEA). Consulter
- Google Analytics Help, Verify and update consent settings in Google Analytics. Consulter
- Axeptio, pricing. Consulter
FAQ
Is advanced mode less compliant than basic mode?
No, both are compliant. In advanced mode tags load but send no cookie and no identifier until consent is granted. The risk is not in the principle, it is in the wiring: a badly deployed advanced mode can send more than intended, so it should be verified rather than avoided on principle.
How many clicks does modelling need?
Google Ads models from 700 ad clicks over seven days, for a single domain and country pairing. The threshold applies market by market: a company operating in three countries can have comfortable total volume without clearing the bar in any of them.
How do I know whether modelling is active on my account?
The most reliable sign is the gap between observed and reported conversions: if no modelled conversions appear, the threshold is usually not being met. Before concluding that the wiring is broken, check your click volume by country over seven days.
Should I change consent platform?
Only if yours does not emit the signals, or emits them after the tag manager loads. That is the thing to check first, and it can be checked without changing anything. Switching platform before running that check often replaces a problem with the same problem.
Can Dotsland help?
Yes. Measuring your real volume by country, arbitrating between the two modes, verifying that the signals fire at the right moment, and working on the banner to lift the acceptance rate. It is the heart of our data, analytics and performance work. Let’s talk, or start with the calculation above: ten minutes in Google Ads tells you which side of the threshold you are on.