Your cookie banner says Reject. Does your ads pixel listen?

Black Friday is on 27 November 2026. If you are running ads that week, the banner on your site has a job: when someone presses Reject, the ads pixel should wait. Here is a two-minute test, what Consent Mode v2 actually does, and the breakages I look for first.

Open laptop on a light wooden desk beside a small potted plant, with a blank screen and a window behind

Black Friday is on 27 November 2026. Before the ads for that week start spending, open your site in a private browser window, press Reject on the cookie banner, and wait ten seconds. Then look at the cookies and the network requests. If a Meta pixel or a Google ads tag has already loaded, the banner was decoration. Look closely. The sale will not fix that, and a busy week, when more pages carry the same pixel and more people hit the banner for the first time, makes the miss both costlier and harder to notice.

This is a website check, not a court opinion. I am not telling you what the law requires in your country, and I am not quoting fines. I am telling you how to see whether the button you show customers does the thing it claims to do, and how that lines up with what Google and Meta have published about consent.

Why is a banner that doesn't block anything a problem?

Because the button is a promise. A visitor who presses Reject has been told the ads cookies will not be used. If the pixel has already sent a PageView, that promise is already broken, and the ads account is collecting events from people who asked it not to. In a sale week the traffic is higher, the pixel sits on more pages, and a broken banner quietly fills the account with events you cannot honestly treat as consented.

It also makes the numbers harder to trust. If some of the events in Ads Manager come from visits that should have been silent, you cannot tell how many of this week's purchases came from people who actually agreed to be measured. That matters when you later change bids or compare this Black Friday with last year's.

The same gap shows up on the landing pages you use for ads. I wrote about keeping one clean page for the sale. That page needs a working banner as much as it needs the right price.

How do you test it in two minutes?

Use a browser you can open in a private window, with the developer tools open. Chrome and Firefox both work. You want to see the site as a new visitor sees it, without your own old consent cookie sitting in the jar, and without any extension that has already answered the banner for you on this computer.

  1. Open a private window and go to your home page, then to the page your ads actually point at. Wait for the banner. If it never appears, that is already the finding.
  2. Press Reject, or the option that refuses ads and analytics cookies. Do not press Accept yet. Wait ten seconds, so late-loading tags have time to arrive.
  3. In the developer tools, open the Application or Storage panel and look at cookies for your domain and for facebook.com, doubleclick.net and google. Then open the Network panel and filter for fbevents, facebook.com/tr, gtag, googletagmanager and google-analytics. Refresh once if the first load was messy, press Reject again, and look at what appears after the click.

You are looking for two different failures. Watch both. One is a cookie that was written anyway, _fbp, _fbc, _gcl_aw or _ga among them, which means the browser stored an ads identifier after the visitor had already pressed Reject. The other is a request that went out anyway, even if no cookie stayed behind. A Meta Pixel PageView after Reject is the one I see first. A Google tag that loads and then says consent is denied is a different case, and I come back to it below.

Then open a second private window, press Accept, and confirm the other side of the test: the pixel should fire now. A banner that blocks everything, including the people who agreed, is also broken. It just breaks the ads account in the opposite direction, and you only notice when the sale is already running and Ads Manager looks empty.

If you advertise on Facebook or Instagram, this test belongs next to the Advantage+ settings check. A pixel that fires for everyone, or for no one, will distort those numbers too.

Google's Consent Mode is a way to tell Google tags whether the visitor has agreed to ads and analytics cookies. The older version had two signals, ad_storage and analytics_storage. Consent Mode v2, which Google asked advertisers to use from 2023 and still documents as the current version, adds two more: ad_user_data and ad_personalization. If you still send only the first pair, Google treats the setup as incomplete for traffic in the European Economic Area, the UK and Switzerland.

The setup has two steps, and both have to happen in the right order. Before anyone touches the banner, the page should set a default, denied for those four signals, and after the visitor chooses, the page should update the signals to match that choice, which is why Google's own documentation describes a wait_for_update window, in milliseconds, for the common case where the banner loads a moment later than the tag. Without that wait, a tag can fire on the default before the banner has had a chance to speak.

What this does not do is just as important. Read this slowly. Consent Mode talks to Google tags. It does not, by itself, pause a Meta pixel, a TikTok pixel, or any other script that was pasted into the theme or fired from a tag manager without its own consent check. Meta documents a separate command: fbq('consent', 'revoke') before fbq('init'), and fbq('consent', 'grant') once the visitor agrees (Meta's consent guide). If that revoke never runs, the pixel behaves as if nobody was asked.

Google also describes two ways to wire this up, in its Tag Manager consent guide. In the basic version, Google tags do not load until the visitor has used the banner. In the advanced version the tags load immediately, send cookieless pings while consent is denied, and send full data only after a grant. Those pings are not the same as a full cookie, and they are also not nothing. I leave the legal reading of that difference to whoever advises you on the law. The practical point is simpler: if you thought Reject meant "no request goes to Google at all", the advanced version is not that.

Google's EU user consent policy is the rule on Google's side for advertisers who measure or personalise ads for people in the EEA, the UK and Switzerland. Consent Mode is how you send the signals. It is not a certificate that the rest of the site is fine.

The one thing I cannot tell you from the outside is whether a particular consent tool is sending those signals correctly. The only reliable check is the one above: Reject, then read the cookies and the requests. A green toggle in a plugin screen is not the same evidence.

Where does this break?

The banner and the pixel are set up by different people, months apart, and the join between them is where things come apart.

The first breakage is a tag manager that loads before the banner has set a default. This is common. The container is in the head of every page, the consent tool arrives later in the footer or from a slow plugin, and for a few hundred milliseconds, before the banner has even drawn its buttons, the default never ran and the first request has already left. Google's wait_for_update exists for this. If nobody set it, or the wait is shorter than the banner takes to load, the first request has already left.

The second is a pixel hard-coded in the theme, in a header script, or in a plugin that does not read the banner's choice. I see this on sites that added a consent tool last year and never removed the old Facebook snippet from header.php, or from a "custom code" box in the site builder. The banner can be perfect and the pixel will still fire, because the two are not connected.

The third is cache. A caching plugin or a server cache stores a copy of the page with the banner already marked as settled, or with the tag already inlined. You press Reject on what looks like a fresh visit and get the behaviour of last week's Accept. The test in a private window after a cache purge is the one that counts; the test you ran on a page you had already accepted on the same computer does not.

A fourth, quieter one is a second domain: a shop on a subdomain, a checkout on a different host, or a landing page built for the sale and pointed at by the ads you are about to run. The home page banner can work and the sale page can still load the pixel immediately. Test the page the ad points at, not only the page you look at every morning. Test that page.

I have made this kind of change on a site this month, with analytics waiting for consent and the ads cookies left off until someone agrees. The test I used is the one above. It took longer to be sure of the sale landing page than of the home page. That is the pattern.

What this is not

The two-minute test is not legal advice, and it is not a promise that a passing result makes the site compliant. Laws and regulators read these questions differently, and a lawyer or a data-protection adviser is the person who answers them. A two-minute test tells you whether the banner and the pixel agree with each other. That is a website fact, and it is one you can fix before 27 November.

It also does not replace conversion tracking. Once consent is granted, the Purchase or lead event still has to fire once, with the right value, on the thank-you page. A silent pixel after Reject is the first job. A working pixel after Accept is the second.

When is it worth getting the site checked?

When you did not build the banner and the pixel yourself, or when the site has had a theme change, a new shop plugin or a new landing page since the last time anybody looked. The check is part of the web design work we do, and it sits next to how we run Google and Meta ads. If you send the address, I will tell you what still fires after Reject and what I would change before the sale.

The useful next step is not a new plugin. It is the private window, the Reject button, and a look at the network panel. If something still leaves after that click, you have a week or two to fix it. After the ads are spending, you will be looking at the wrong numbers and calling it a sale problem.

Read next