ANALYTICS
Consent Mode v2 one year on: the actual impact on conversion data.
Google enforced Consent Mode v2 for EEA traffic on 6 March 2024. Without it, personalised advertising data from EEA users stops flowing to Google Ads. One year on, here is what the data looks like across four EEA-heavy clients.
Setup
Four clients (two DTC, one B2B SaaS, one travel) with >60% EEA traffic. All four migrated to Consent Mode v2 (advanced mode) between Feb and April 2024. We compared conversion reporting continuity in Google Ads, GA4, and warehouse (source of truth) across three windows: pre-migration (Nov 2023 to Jan 2024), transition (Feb to May 2024), post-migration (June 2024 to June 2025).
What moved
Consent rate: Across the four, consent rates landed at 52%, 68%, 71%, and 44% (DTC, DTC, B2B, travel). The travel vertical is lowest because of cross-border cookie banners from third-party booking widgets confusing users.
GA4 modelled conversions: In advanced mode, GA4 models conversions for consent-denied traffic. For the four clients, modelled conversions made up 11 to 24% of total reported conversions. We validated against warehouse data. Modelled conversions were within ±6% of the warehouse truth on three of the four. The fourth (travel) had a 14% overcount.
Google Ads smart bidding: Campaigns restored to within a few percent of pre-2024 performance after modelled conversions started flowing. Teams that stayed on basic Consent Mode saw 20 to 40% reported-conversion deficit through the year, and their smart bidding underperformed accordingly.
What did not move
Consent rate did not improve over the year despite UX tweaks on banners. Teams that rebuilt their banner copy 3 times saw consent rates drift ±3%, within noise.
Server-side CAPI to Meta was not affected by Consent Mode v2 at all. That is a Meta story, not a Google one.
Lessons
Advanced mode is meaningfully better than basic mode for ad platform performance. Basic mode lost enough data that smart bidding went off the rails.
GA4 modelling is accurate enough for most decisions if consent rate is above 50%. Below that, the model gets noisy.
The conversion gap from denied consent is a warehouse problem. Get your warehouse logging consented and unconsented events (server-side, aggregated) so you have a source of truth that is not the ad platform.
Sources: four client implementations and audits, Feb 2024 to June 2025; Consent Mode v2 documentation; IAB Europe TCF v2.2 updates.