How a consent update works
Consent Mode has two moments. Before anyone touches the banner, the page sets a default consent state — normally denied for the storage types that need permission. Tags still load, but they behave in a restricted way. Then the visitor clicks accept or reject, and the consent platform fires the update: a message telling every tag on the page which storage types are now granted and which stay denied.
The update carries named parameters — analytics storage, ad storage, ad user data and ad personalisation — each switched to granted or denied. A tag that was writing nothing to cookies starts writing them; a tag that was sending stripped-down pings starts sending complete ones. Nothing is retrospective. Hits already sent under the default state stay exactly as they were sent.
Why consent updates matter
Without one, the choice a visitor makes never reaches your measurement stack. The banner disappears, the visitor assumes something happened, and the tags carry on in whatever mode they were defaulted into. You end up with the worst of both outcomes: either you look compliant while measuring far less than you are allowed to, or you look compliant while still collecting from someone who said no.
Timing matters as much as presence. If the update arrives after the first page view has already been sent, that hit is recorded under the default, so part of your analytics sits in the restricted bucket even for people who accepted. Across a full month of sessions the reporting drifts quietly away from reality, and the gap stays invisible unless somebody goes looking for it.
Where consent updates go wrong
The most common fault is a banner that stores the choice but never broadcasts it. The consent platform writes its own cookie, the site remembers the visitor next time, but no update is pushed to the data layer, so the measurement tags never learn what was clicked. The second is the returning visitor: the choice was made on an earlier visit, so the platform has to replay the update early on every later page load. If it does not, a consenting regular is treated as a refusal for ever.
A third fault is legal rather than technical — an update that grants everything regardless of which button was pressed, usually copied from a demonstration snippet and never edited. It makes the reports look healthy and the consent record worthless at the same time.
How to act on it
Test in a real browser rather than trusting the consent platform’s own dashboard. Load a page with tag debugging open, watch the consent state before the banner is touched, then click accept and watch it change. Repeat the whole test with reject, which is the path nobody checks and the one a regulator would look at first.
Then clear the browser, accept, close it and come back, to confirm the stored choice is replayed on load rather than asked again. If any of that fails, the fix belongs with whoever owns the tag setup — this is exactly the kind of silent gap a tracking audit is meant to surface before it corrupts months of reporting.
