TagAudit's free scan opens the page you enter with no stored consent, presses your banner's reject button when it can find one, and records which analytics and marketing tags it sees still sending requests. A separate pass tries to accept, for comparison. Where no reject control is found, the report says so and makes no claim about whether consent is respected.
What we check
| Check | What the report shows |
|---|---|
| Tags sending requests before any consent choice | Each tool's consent column: "Sent requests before any consent choice was made" |
| Ad tags still trying to send data after Reject was pressed | A HIGH finding, raised only when a reject control was actually clicked |
| Tags seen both with consent withheld and after accepting | A separate finding |
| Consent Mode | Whether tags sent a Consent Mode ping announcing the denial |
| Which consent platform is on the page | 15 named platforms recognised, plus an unidentified-banner match |
Sample finding (how a finding reads; not data from a real site): HIGH · Ad/marketing tags fire despite declined consent
Does my cookie banner actually block tags?
Only a test can tell. A banner can be installed correctly and still let half the stack through. The scan presses Reject when it finds the button, then lists the tags it saw still sending requests, so you see the gap instead of assuming it.
What if the scan cannot find a Reject button?
Then it does not guess. The report labels the session "Withheld (no reject control found)" and says it makes no claim about whether consent is respected, in either direction. If your banner really has no reject option, that is worth knowing on its own.
What this can't tell you
- Whether the site is GDPR compliant. The scan gives technical evidence, not a legal verdict.
- What happens on a banner it cannot operate. Reject is pressed only where a control is found: built-in support covers five major consent platforms, plus common reject-button text and ids.
- Server-to-server data. Meta CAPI and offline uploads aren't visible from outside.