Google Consent Mode adjusts how Google’s tags behave based on what a visitor agreed to, and where consent is refused it can model the conversions you lost. That modelling has a volume requirement, and the requirement excludes a large share of the advertisers who would benefit most.
The number is 700 ad clicks over seven days, split by country. Below it, you lose the data and receive nothing in exchange.
This is for you if you run a consent banner and have never checked whether the modelling behind it is actually running.
What Google Consent Mode is doing
It replaces a blunt outcome with a graded one.
Without it, a refused consent usually means your tags do not fire. Google learns nothing, the conversion is invisible, and your reporting simply undercounts.
With it, your site sends signals describing what was and was not agreed. Google’s tags adjust rather than disappear, sending cookieless pings that carry no identifying data. Those pings are what the modelling later works from.
So the mechanism is sensible and the implementation is mostly a tag manager and banner configuration job. The catch is not in the setup. It is in who qualifies for the benefit.
The 700 click threshold, and why it is stricter than it sounds
Google’s documentation states the requirement plainly: 700 ad clicks over a 7 day period, per country and domain grouping.
Read the qualifier carefully, because that is where the difficulty lives. It is not 700 clicks across your account. It is 700 per country, per domain grouping.
An advertiser running in the UK, Ireland, Germany and France needs to clear 700 in each of those separately. Spreading a decent budget across four markets can mean failing the threshold everywhere while looking substantial in aggregate.
Seven days is also a short window at low volume. A modest lead generation advertiser spending a few thousand a month in one country may sit under it permanently, with no indication in the interface that this is happening.
Who Google Consent Mode actually helps
Large advertisers, concentrated in few markets. That is the honest answer.
If you run high volume in one country, you clear 700 clicks easily, modelling engages, and much of the consent-refused loss comes back. The mechanism works as designed.
If you run modest volume, or spread it across several markets, you sit below the threshold. You still lose conversions to refused consent, and you get no modelling to offset it.
So the cost of consent falls on everybody and the remedy is available to some. I have not seen that argument made anywhere, and it seems worth stating, because it changes what a smaller advertiser should expect from an implementation.
It also changes the honest sales pitch. Setting this up for a small advertiser is worth doing for correct tag behaviour. Promising recovered conversions is promising something the threshold may prevent.
The banner configuration that undermines it
There is a common setup that quietly defeats the whole thing, and Google warns about it directly.
Google’s guidance says that if you prevent your Google tags from loading until a user interacts with your consent banner, Google will not be able to verify user consent choices, and this may lead to loss in data.
That is a very common banner configuration. Plenty of implementations block everything until somebody clicks, which feels like the cautious choice and removes Google’s ability to see the consent state at all.
The correct pattern is to load the tags in a default denied state, then update when the visitor chooses. Tags present, behaving conservatively, rather than absent.
Worth checking on your own site, because this is usually set once by whoever installed the banner and never revisited.
What changed around Google Consent Mode in 2026
One dated fact belongs here, because it caught people out and it is easy to miss.
From 15 June 2026, the advertising storage permission became the sole gate on advertising data reaching Google Ads from GA4. Previously other signals could carry some of that load. After the change, if advertising storage is denied, the data does not flow, regardless of what else your setup does.
That is reported consistently by several consent platform companies. I could not find Google’s own announcement page for it, so attribute it to the platforms rather than to Google directly, which is exactly the sort of qualification this topic usually skips.
The practical effect is that banner categories matter more than they used to. A visitor who accepts analytics but declines advertising now removes themselves from Google Ads reporting entirely, and plenty of banners present those as separate choices.
So if your consent rates look healthy overall but your Google Ads conversions dropped in mid-2026, check the split between categories rather than the headline acceptance figure.
What the modelled conversions look like in your reports
They arrive without a label, which affects how you should read recent data.
Modelled conversions appear in the standard Conversions column alongside observed ones. Nothing marks them as estimates. So a figure you quote on Monday contains a mixture you cannot separate from the interface.
They also take up to five days to stabilise. Recent performance therefore moves after you have looked at it, and a campaign that appeared weak on day two can look different on day six without anything changing in the account.
The practical rule is to avoid judging the last five days at all where consent modelling is active, and to say so when somebody asks why the numbers moved.
What nobody has measured
There is no independent study of what Google Consent Mode recovers in practice.
What exists is vendor marketing, mostly from consent platform companies, and scattered anecdotes. I looked for a controlled comparison with a disclosed sample and found none.
On European consent rates themselves there is at least a dataset, though it needs its caveat spoken aloud. Didomi’s benchmark puts Southern Europe at 82.5% consent, with Eastern Europe highest at 89.3% and Western Europe lowest at 75.1%. Didomi is a consent platform, the sample is its own customers who have invested in consent design, and no sample size is disclosed.
So the best available figures come from an interested party with an undisclosed sample. That is worth knowing before anybody builds a forecast on them.
What to change this week
Three steps.
Work out whether you clear 700 ad clicks in seven days in each country you advertise in. That single check tells you whether modelling is doing anything for you at all.
Then look at how your consent banner loads tags. If everything is blocked until somebody interacts, you have the configuration Google warns against.
Then stop judging the most recent five days of conversion data where modelling is active, and tell whoever reads your reports why those numbers move.
What conversion numbers actually measure covers the counting underneath, third party cookies explains what did and did not change in the browser, server side tracking is the architecture question that sits beside this, and why your reported cost per lead is wrong covers the reporting consequences. Banner and tag behaviour is usually a web development job rather than an advertising one. To find out whether your modelling is actually running, book a teardown.