Consent Mode v2: what effect does it have on your Google Ads campaigns?

Consent Mode v2 has been required by Google since March 2024 for any advertiser serving to users in the European Economic Area. Its principle and its technical configuration are described in Consent Mode: definition and Consent Mode v2 with GA4 and GTM. This article deals with what actually happens in a Google Ads account depending on whether the set-up is missing, badly implemented or correctly deployed, and with what can be recovered.

What changes in the account

Conversion modelling and its thresholds

When a visitor refuses cookies, the link between their ad click and their purchase disappears. Google fills that gap statistically, from the behaviour of consenting visitors, and puts the estimated conversions back into the reports. This modelling is only triggered above certain volumes, of the order of a thousand ad clicks per month per country, and several hundred observed conversions over a week. Below that, an SME account gets nothing: the conversions from refusals are simply lost, and the cost per acquisition displayed is overstated by the same amount. That is one more reason to read the results in GA4 or in the CRM rather than in the ad interface alone, as pointed out in attribution.

The effect on automated bidding

Smart Bidding decides each bid from the conversions it observes. If 30% of conversions do not come through, the algorithm underestimates the value of certain campaigns, in particular those reaching audiences on privacy-protective browsers or more inclined to refuse cookies. The consequence is a gradual reallocation of budget towards the best-measured campaigns, which are not necessarily the most profitable. The mechanism is described in Smart Bidding. Two fixes limit this effect: enhanced conversions, which send Google hashed contact data collected with consent, and importing offline conversions from the CRM.

The checks to run

1. The Google Ads diagnostics: the conversion diagnostics section and the account warnings flag missing signals ad_user_data and ad_personalization.
2. Tag Manager preview: check that the four signals are sent with the expected state before and after the visitor's choice.
3. The consent rate: readable in the dashboard of the CMP ; below 50%, the banner itself needs reworking.
4. The share of modelled conversions: visible in GA4 and in the Google Ads reports; a zero share despite substantial traffic points to a badly implemented set-up.
5. Audience size: a remarketing list shrinking month after month confirms that the signals are not getting through.
6. The gap with the back office: compare the conversions reported with the real orders or quotes; that is the check that settles it.

Recovering what can be recovered

• Improve the consent rate: the most direct lever; ten points gained make ten per cent more conversions visible.
• Turn on enhanced conversions: sending hashed contact data at the time of conversion, which improves matching.
• Import offline conversions: deals signed and fed back from the CRM are worth more than modelling, especially in B2B.
• Move to server-side tracking: the same logic as the conversions API at Meta, described in CAPI.
• Accept imperfect measurement: steering by cost per signed deal rather than by cost per reported conversion remains the most robust solution for an SME.

The frequent mistakes

• Installing an uncertified CMP: the signals are not sent in the expected format, and the account stays in warning.
• Blocking the tags entirely: in basic mode, no modelling is possible; it is a defensible choice but it must be a deliberate one.
• Believing the set-up restores measurement: it recovers part of it, never all of it.
• Not retesting after a redesign: a change of site or of CMP frequently breaks the transmission of the signals.