Quick Answer
A cookie banner should be reviewed together with the scripts, storage and data transfers it is meant to control. Inventory the site’s providers, define expected behavior from an approved privacy policy and legal advice, then test fresh visits, choices, withdrawal and return visits. A visible banner or a denied setting alone does not establish that every integration follows the intended rules.
Separate the policy decision from the implementation evidence. Qualified advice should determine applicable requirements, permitted purposes and appropriate notices. A technical review can then check whether the website behaves as specified.
Inventory what runs and what it sends
List analytics, advertising, embedded media, chat, forms, payment tools, personalization and other integrations. Record who adds each one: theme code, plugin, tag manager, app or an external service. Include server-side or platform events that a browser-only scan cannot observe completely.
For each provider, record its purpose, destination domains, browser storage, relevant identifiers and documented consent controls. Map the implementation to an approved category and owner. An unfamiliar request should become an investigation item rather than being labeled necessary or optional by guesswork.
Cookies are only part of the inventory. Review other browser storage and requests even when no new cookie appears. Keep raw evidence protected when it contains identifiers or URLs with personal information.
Write the expected behavior before testing
Record the approved defaults, categories, regional behavior, preference duration and withdrawal route. Ask qualified privacy advice to resolve the legal classification of purposes, applicable opt-out signals and notice wording.
Connect each decision to an implementation control. Identify the code or provider setting that prevents, delays or changes a request. A consent platform’s category label does not prove a separately installed script uses that category.
For Google tags, the official consent mode overview distinguishes basic blocking from advanced mode, which can send measurements without cookies while consent is denied. Choose the implementation according to approved requirements; do not interpret “denied” as a universal promise of zero requests.
Use a consent-state test matrix
| Scenario | Evidence to compare with the approved behavior |
|---|---|
| Fresh visit, no interaction | Initial state, script execution, requests and storage before a choice |
| Reject or deny optional purposes | Updated states and behavior of each mapped provider |
| Accept selected categories | Only the approved category changes and their dependencies |
| Withdraw a previous choice | Updated controls, subsequent requests and documented storage handling |
| Reload or return | Whether the saved preference is restored before dependent behavior |
| Applicable privacy signal or region | The configured test condition and the resulting behavior |
Use a controlled test context and record the browser, relevant region configuration, provider versions, page and time. Repeat on representative page types, including pages that add media, forms or commerce integrations absent from the homepage.
If Global Privacy Control is part of the approved requirements, test an environment that sends the signal and verify the implementation’s response. MDN documents the browser property and its support limits. An absent or false property should not be treated as affirmative permission.
Check timing and provider-specific controls
Inspect whether defaults are established before dependent tags run and whether updates reach them before navigation. Google’s implementation guide explains that ordering. Other vendors need their own documented integration and verification.
Check for duplicate tags installed through both code and a manager. A controlled copy can reveal whether an apparently blocked provider still loads through another route. Browser inspection alone may not establish server-side behavior, downstream retention or every data use; record those limits and ask the provider for the missing evidence.
Review controls and keep the evidence current
Test keyboard access, readable labels, focus visibility, narrow layouts and the route for reopening preferences. Compare the displayed choice with the actual stored and transmitted state. A usable preference control and a correctly mapped integration are separate acceptance checks.
Record each finding with the expected rule, observed evidence, affected provider, owner and retest. Revisit the inventory after new apps, pixels or embedded tools are added. Our privacy and consent implementation service can support this technical work alongside the business’s qualified legal review.
Key Takeaways
- Inventory providers, requests and storage before judging the banner.
- Turn approved policy decisions into testable technical expectations.
- Test fresh, denied, selected, withdrawn and returning states.
- Record provider and server-side evidence gaps without claiming certification.
Frequently Asked Questions
Does installing a banner make a website compliant?
That cannot be established from installation alone. Applicable legal requirements and actual implementation both require review within the business’s circumstances.
Does denied consent mean no Google requests?
Not in every implementation. Google’s advanced consent mode can send measurements without cookies. Basic and advanced modes behave differently.
Can a scan find every privacy issue?
No. It observes a particular context and may miss interactions, return visits, region differences, server-side events and downstream uses. Keep its scope and limits with the result.

