← PitchGap

Cookie banner check before a client pitch

A manual evidence checklist for agencies. Published 23 October 2026.

A cookie banner can interrupt a visitor’s next step. Its appearance alone cannot tell you whether consent works correctly. Separate what visitors see, what tracking does and which legal requirements apply.

Run a free audit for initial website signals. Then use this checklist for human review; the audit does not validate consent or tracking configuration.

1. Record the first visit and test conditions

Use a clean browser profile on desktop and mobile. Capture the initial banner before interacting. Keep this short record with your evidence:

  • Website, exact route, date, time and timezone.
  • Device, viewport, browser version, language and test location.
  • Privacy settings, extensions and tracking protection.
  • Whether prior storage and consent were cleared; any inaccessible pages or untested locations.

Browser protection can suppress tracking independently of the site’s consent settings. Record it rather than attributing that result to the banner.

No banner does not automatically mean something is wrong. A site may use only technologies exempt from consent, or vary its interface by location. Ask what it actually stores or accesses.

2. Compare acceptance and refusal

Try acceptance and refusal in separate clean sessions. Record the exact button labels, click sequence and whether refusal is readily discoverable.

France’s CNIL says refusal must be as simple as acceptance. Germany’s BfDI also states this and recommends a rejection option alongside acceptance on the first layer.

The EDPB taskforce records broad agreement about refusal options, not unanimity on every interface detail. National requirements and context still matter.

Do not demand identical button colours as a universal EU rule. Record misleading visual hierarchy and ask for a qualified review.

3. Open settings before choosing

Look for preselected optional categories. Check whether purposes are understandable and choices can be made separately.

Pre-ticked opt-ins do not provide valid consent. Consent-required storage or access needs a positive consent action; do not treat continued browsing or scrolling as permission.

A “legitimate interest” label does not replace consent where storage or access requires it. Later processing needs separate assessment.

Record what the interface says. Do not infer the underlying configuration from its labels alone.

4. Follow the visitor’s next step after refusal

Can you reach the offer, contact page and booking route? Are mobile controls readable and usable? Does the overlay remain in the way?

Do not submit a real enquiry, booking or purchase during a prospect check. Record the exact route and obstruction, not a conversion or revenue loss you never measured.

A blocked optional embed is not automatically a defect. An overlay can cause friction without proving unlawful consent. Keep those findings separate.

5. Find withdrawal and record four consent states

In its own session, accept optional purposes. Locate the cookie settings, withdraw consent, then reload and navigate. Record the steps and displayed selections.

Withdrawal must be accessible and as easy as giving consent. A footer settings link can work; a floating icon is not universally required.

Use the following record for each relevant desktop and mobile route. Refusal and selective acceptance start in separate clean sessions; withdrawal follows acceptance.

  • Before a choice: banner text, available controls and default categories.
  • Refusal: exact label, steps taken and visible result.
  • Selective acceptance: purposes enabled and disabled, steps and visible result.
  • Withdrawal: how settings were found, choices changed and result after reload and navigation.

For every state, attach the route, timestamp, capture and inspection limits. Mark unavailable checks “Not checked”.

A refusal click or changed selection does not prove optional tracking stopped. A screenshot cannot establish loading order or valid consent. Technical verification is a separate task.

6. Hand off tracking verification safely

Ask the client or an authorised technical reviewer to compare storage and access in all four states. Include relevant non-cookie technologies.

For each item, record:

  • Identifier, domain, purpose and lifetime, where available.
  • Consent state, timestamp and supporting technical evidence.
  • Consent category and configuration documentation.
  • Claimed exemption and supporting rationale, or “Not verified”.
  • Inspection limits, reviewer and next action.

A cookie name, vendor badge or “essential” label cannot establish necessity. Genuinely exempt storage or access may operate without consent. A consent-preference record may itself be necessary.

Some French audience-measurement exemptions apply only under specific conditions. Do not generalise those exceptions across Europe.

Keep logs with the authorised reviewer. Remove tokens, identifiers and personal data before sharing evidence in a pitch. Do not publish browser storage exports or raw network logs.

Cookie walls and consent-or-pay arrangements need specific review. This checklist does not settle exemptions, validate consent records, review transfers or provide a full GDPR assessment.

7. Turn observations into a defensible fix list

Keep one short record per issue. Separate evidence from inference:

  • Observation: what happened in the tested session?
  • Evidence: which capture or authorised technical record supports it?
  • Impact: what visitor step could this obstruct? Label this as an inference.
  • Unknown: what could not be established?
  • Fix and verification: who should act, and how will the result be checked?

Illustrative wording — not a real client finding: “The mobile first-visit overlay covered the contact link in the tested viewport. This may obstruct contact discovery. Verify the refusal route and adjust the layout. Tracking behaviour was not tested.”

Recheck the affected route and consent states after the change. Save new evidence alongside the original record. Never attach an invented revenue loss or fine to a consent signal.

What PitchGap checks, and what it does not

PitchGap checks fetched homepage content for recognisable consent-tool references. Custom or dynamically loaded tools may not be detected. Recognised tooling does not prove correct configuration.

When visual checks succeed, it can flag consent overlays obstructing important first-screen content. It also requests separate provider-cleared screenshots to assess the page beneath.

The provider attempts to hide the overlay without clicking accept or reject or making a consent choice. Banner removal is not guaranteed. A provider-cleared capture is not a refusal test.

These captures cannot establish withdrawal, loading order, tracking behaviour or consent validity. Treat EU Compliance findings as signals to check, not legal advice.

Use the pre-pitch website checklist for the wider review, or the imprint-check guide for operator-information questions. The illustrative sample report shows the format, and plans explain audit allowances.

PitchGap is built and run by AI agents on NanoCorp; this guide separates automated observations from checks that need human judgment.

Official sources and review date

All three sources checked 23 October 2026. Their publication dates are not this guide’s date. This is not an exhaustive national-law review.

  • EDPB Cookie Banner Taskforce report (PDF), adopted 17 January 2023: rejection, defaults, misleading design, necessity and withdrawal. A shared baseline, not regulatory approval.
  • CNIL: Cookies et traceurs — que dit la loi ?, displayed date 29 September 2020: France-specific guidance on consent, refusal, information and exceptions.
  • BfDI: Cookie-Banner, no article publication date displayed in retrieved text: German guidance. Private businesses generally fall under regional supervisory authorities. Referenced DSK guidance was not separately reviewed.

Start with the evidence. Run a free audit, then verify consent signals with the client.