searchDecoded Labs · Measurement explainer

GA4 consent mode simulator

See what happens to Analytics measurement when visitors don't consent to analytics cookies, and why a drop in observed traffic doesn't always mean fewer people visited.

The question

What can Analytics actually observe?

Your website can receive a visit even when Google Analytics 4 (GA4) can't observe it in the conventional cookie-based way. This simulator separates estimated website activity from the share that grants analytics consent.

It is a simplified educational model, not an audit of your consent banner or GA4 property. It can't calculate your true reporting difference or predict how much behavioural modelling may add.

Build a scenario

Follow the audience through measurement

Change the hypothetical traffic, consent rate and implementation. The audience remains the same. Only the way its activity can be measured changes.

Start with

Enter a hypothetical total from 100 to 100 billion visits.

Measurement implementation

Hypothetical website activity

100,000monthly visits

Conventionally observable

70,000consented visits

Potential measurement gap

30,000visits without analytics storage

100 representative signals

One dot represents 1% of the hypothetical audience

ConventionalWithout analytics storage

70 of 100 signals represent conventional measurement. The other 30 represent activity without analytics storage, for which measurements without cookies may be sent as potential modelling input.

Selected setup

Advanced Consent Mode

With analytics consent, GA4 can use analytics cookies for conventional measurement. Without consent, analytics cookies are not read or written, but measurements without cookies may still be sent.

These cookieless measurements can be an input to behavioural modelling only when Google determines that the property meets its eligibility and model-quality requirements. No recovery amount can be predicted here.

Consent-rate sensitivity

What a ten-point change would mean

The website activity stays fixed. This compares conventional observability at the current consent rate and ten percentage points either side.

60%
Conventionally observable
70,000
Potential gap
30,000
70%Current
Conventionally observable
70,000
Potential gap
30,000
80%
Conventionally observable
70,000
Potential gap
30,000

Figures use the same monthly website activity as the active scenario. They do not predict a future consent rate.

Compare the setups

The consent gap stays the same, but the signal changes

All three examples respect the visitor's choice. The difference is whether Google tags remain blocked or adapt their behaviour after analytics storage is denied.

01

Tags blocked / no Consent Mode

Consent granted: Analytics measurement may be available.

Consent denied: Blocked analytics tags send no GA4 measurement from those visits.

02

Basic Consent Mode

Consent granted: Google tags load and conventional measurement is available.

Consent denied: Google tags stay blocked and no data is sent from the blocked tags.

03

Advanced Consent Mode

Consent granted: Analytics storage is granted and conventional measurement is available.

Consent denied: Analytics cookies aren't read or written. Measurements without cookies may be sent for basic measurement and modelling.

Understand the signals

What Consent Mode v2 actually controls

Consent Mode does not reduce every choice to one yes or no. Google tags receive a granted or denied state for separate purposes. Your consent banner or consent management platform is responsible for sending states that match the visitor's choices.

When analytics storage is denied

A measurement without cookies is not normal cookie-based measurement

With an advanced implementation, Google Analytics does not read or write analytics cookies when analytics_storage is denied. A limited measurement may still communicate the consent state and activity to Google for basic measurement and modelling.

This does not change the visitor's choice, grant analytics storage or recreate a named person. It gives Google less information than conventional measurement with analytics cookies.

Keep the concepts separate

Behavioural modelling and key-event modelling answer different questions

Behavioural modelling

Estimates user and session behaviour that GA4 cannot directly observe when identifiers such as analytics cookies are unavailable. Google trains the model using observed data from the same property. It does not reconstruct individual visitors.

Key-event modelling

Estimates important actions that cannot be observed directly. A key event is an action marked as important in Analytics, such as a completed enquiry or purchase. This modelling supports reporting and attribution, but it is not the user-and-session estimate shown by behavioural modelling.

Behavioural modelling prerequisites

Check the published GA4 thresholds

Compare a property with Google's published prerequisites for behavioural modelling. Meeting every item does not guarantee that Google will activate modelling because separate model-quality requirements still apply.

Measurement implementation for this check

Preset from the main simulator until you choose a different option here.

Average events per day sent with analytics_storage='denied'.

Google's published prerequisite is at least seven days.

Average daily users sending events with analytics_storage='granted'.

Count the qualifying days within the previous 28 days.

Method and limits

What the simulator calculates

The conventional estimate is monthly visits multiplied by the analytics consent rate. The potential measurement gap is the remaining share. Both figures are rounded to whole visits.

The calculation assumes the supplied consent rate applies evenly to the hypothetical audience. It doesn't account for repeat visitors, sessions, events, consent-banner behaviour, tag errors, regional rules, reporting identity or other causes of GA4 differences.

Advanced Consent Mode never adds a modelled number. The eligibility checker compares the selected implementation and supplied figures with Google's published minimum prerequisites. It cannot test Google's separate model-quality criteria or confirm that modelling will activate.

Embed this simulatorAdd the current scenario to another website.

Keep exploring

Measurement is not the same as activity

If reported traffic changes after a consent update, compare the implementation, consent rate and tag behaviour before treating the change as a fall in real visits.

Explore all Labs