Conversions dropped after Consent Mode v2: how much of a drop is normal?

Some drop is expected, because people who decline are no longer measured the old way. A large drop that starts on one date and does not match your decline rate usually points to a setup problem. The way to tell is to compare against something that does not depend on consent, like your orders or CRM.

First, anchor on a source of truth. Your store orders or your CRM count real sales regardless of cookies. Put them next to the platform's reported conversions for the same days. If real sales are flat and reported conversions fell, the loss is in measurement, not demand.

Second, look at timing. Did the fall start on the day the banner or Consent Mode went live? A sharp step on that date suggests a configuration cause. A slow slide that began earlier suggests something else.

Third, estimate what you should expect. If, say, a bit under half of visitors accept, the measured share of conversions will fall toward that number, partly offset by Google's modelling where it applies. The exact effect depends on your region, your traffic mix and the mode you run, so I would not trust any rule of thumb you find online. Use your own acceptance rate as the reference point.

Fourth, check the usual faults. Consent defaults set after Google tags fire, a banner that never updates the signals after the visitor accepts, a tag that requires a signal your banner does not send, or a conversion tag blocked from firing until a second page load. Each of these is testable on a clean browser.

Fifth, check that accepted visitors are measured. Accept on a test visit, make a test conversion, and confirm it appears. If accepted visitors are also missing, the problem is the wiring, not consent itself.

Sixth, consider server-side events for the consented group. When a visitor accepts, a server copy of the conversion with the click ID or hashed identity can fill gaps that the browser misses, if your rules allow it. For visitors who decline, nothing identifiable should be sent.

DataCops sends consented conversions from the server with the same event ID as the pixel, so each counts once, and checks the saved choice before every send, logging skips. That helps you tell 'declined, correctly skipped' from 'accepted, never sent'. It does not make up data for visitors who said no, which is the point.

What is your banner's acceptance rate, and did the drop start on the go-live date?