What does Google Consent Mode v2 do when a shopper rejects consent?
Denied consent does not mean zero measurement. Consent Mode v2 sends cookieless pings that Google models into aggregate conversions, and the setup details decide how much signal survives.
Rejection is not the end of measurement
When a shopper clicks "reject all" on a consent banner, many store owners assume Google Ads and Analytics go blind for that session. Under Consent Mode v2, that is not what happens. The tags still fire, but in a restricted mode: cookieless pings carry no identifiers and no personal data, and Google uses them to model aggregate conversions. The store keeps directional measurement instead of a black hole. How much signal survives depends on setup details most stores never check.
What actually gets sent on denial
With ad_storage and analytics_storage denied, Google tags send pings that include only coarse, non-identifying information: that a page was viewed, that a conversion happened, rough timing, and limited campaign context. No cookies are set or read, no device identifiers are attached, and the data cannot be joined to a user profile. Google aggregates these pings across many denied sessions and applies its modeling to estimate the conversions that would otherwise be invisible.
This is explicitly not a workaround for consent. The modeling only works in aggregate, it only works when enough denied-session volume exists, and it degrades gracefully rather than pretending to precision. A store that treats modeled conversions as exact counts will misread its data. Treat them as what they are: better than nothing, worse than consented measurement.
The implementation details that decide everything
Modeled data only flows when the implementation is correct, and most broken implementations fail silently. The common defects: tags firing before the consent state is known (so the denial never registers), the default consent state set to granted instead of denied (which is both non-compliant and corrupts the modeling), consent signals not passed to every tag on the page, and the banner's choice not persisting across pageviews so the mode flaps mid-session.
Verify with the browser, not the dashboard. Open the store in a fresh session, reject consent, and watch the network requests: the pings should carry the denied state flags. Then accept in another session and confirm the flags flip. If the flags do not match the banner choice, the modeling is built on lies and the reports are fiction.
Advanced vs basic mode, in plain terms
Consent Mode v2 comes in two flavors. In basic mode, Google tags are fully blocked until consent is granted; denied sessions send nothing at all, and there is no modeling for them. In advanced mode, the cookieless pings described above flow on denial. The tradeoff is real: advanced mode preserves more measurement but requires careful configuration and a clear-eyed view of what the pings contain. Stores that chose basic mode during setup and forgot about it are leaving modeled conversions on the table.
Which mode is right depends on the store's risk posture and its regulators, but the decision should be deliberate, not accidental. Check which mode is actually deployed. Many stores believe they run advanced mode because their CMP advertised it, while the tag configuration tells a different story.
What denial does to your ad bidding
Modeled conversions feed back into Google Ads bidding, but with a lag and with less granularity than consented data. Campaigns in regions with high denial rates (much of the EU) will see noisier performance signals, and smart bidding needs more volume to stabilize. Practical responses: consolidate campaigns to give the algorithms more signal per campaign, lengthen evaluation windows before judging performance, and stop comparing denied-heavy periods to consented baselines as if they were measured the same way.
Also mind the interaction with Enhanced Conversions. Enhanced Conversions hashes first-party data the shopper provided (like an email at checkout) and can supplement measurement, but it has its own consent requirements. It is not a back door around a denial. Configure it for the consented path and leave the denied path to the cookieless pings.
The honest summary for stakeholders
When the marketing lead asks what a 40 percent reject rate means, the answer is: we measure those sessions in aggregate through modeling, with less precision and a delay, and our optimization needs more volume and patience than it used to. That is a defensible position. "We have no idea" is not, and neither is pretending the modeled numbers are exact. Consent Mode v2 done right turns rejection from a measurement catastrophe into a measurement handicap. Done wrong, it is a compliance risk wearing a measurement costume.