Why Google Analytics traffic drops after adding a cookie banner

You installed a cookie consent banner last week. Within days, your Google Analytics reports show 40% fewer users. Your marketing team is panicking. A client wants answers. The banner must be broken, right?

It is not broken. Your actual website traffic has not changed at all. The same number of people are visiting your pages, clicking your links, and browsing your products. What changed is how many of those visits get recorded in GA4.

A properly configured cookie banner blocks analytics scripts until the visitor explicitly clicks Accept. Under EU rules, that consent must come before any tracking cookie is set. Many visitors never give it: they click Reject, ignore the banner, or leave before interacting with it. Those visitors become invisible to Google Analytics, so the drop you see is the share of visits that did not consent.

The sections below cover how to size the drop for your own site, how to check that real traffic is unchanged, and what Google Consent Mode can and cannot bring back.

How a cookie banner causes a traffic drop in GA4

Before the banner, Google Analytics loaded on every single page view. The _ga and _ga_<container-id> cookies, the only two that Google lists for GA4, were set the moment a visitor arrived. Every session, every page view, every event was captured. Your reports reflected close to 100% of actual traffic.

After the banner, the rules change. Under the ePrivacy Directive and the GDPR, analytics cookies like _ga are classified as non-essential. They require explicit, informed consent before being placed on the visitor's device. Your cookie banner enforces this by blocking the GA4 script until the visitor clicks Accept.

Visitors who reject cookies, close the banner, or scroll past it without interacting are never tracked. GA4 does not know they exist.

This is not a bug. This is the banner doing exactly what it should.

The drop is most visible in three areas of your GA4 reports:

• Real-time reports show an immediate decrease the day the banner goes live

• User and session counts fall compared to the previous period, by roughly the share of visits that did not consent

• Conversion attribution suffers because fewer journeys are tracked end-to-end

Meanwhile, these metrics remain completely unaffected:

• Google Search Console clicks and impressions (independent of cookies)

• Server access logs

• Hosting provider analytics dashboards

Why Google Analytics stopped tracking some of your traffic

If GA4 stopped tracking part of your traffic on the day a banner went live, the banner is almost always the reason, and the missing part is the visits that did not accept analytics. Look for a step change on that exact date rather than a slow decline. A slow decline, or a drop in regions where you show no banner, points to something else: a broken tag, a theme update that removed the snippet, or a real fall in visits.

What consent rates actually look like

The size of your traffic drop depends on your consent acceptance rate, which varies by geography, industry, and banner design. No single published figure applies to every site, so the most reliable number is your own: your consent tool's analytics show how many visitors accept, reject or ignore the banner.

A quick calculation with hypothetical numbers: if 60% of your EU visitors accept cookies and 80% of your total traffic comes from the EU, you will see roughly a 32% drop in GA4-reported traffic. That is arithmetic, not a malfunction.

Google Consent Mode v2 and the missing data

This is where Google Consent Mode v2 changes the picture, for sites large enough to use it. It cannot bring back the visits themselves, but it can add estimates for them.

With Consent Mode in advanced mode, the GA4 tag loads on every page visit, even before the visitor interacts with the banner. But it operates in a restricted state: no cookies are set and no client identifier is sent. Instead, GA4 sends cookieless pings to Google's servers.

Google uses these pings, combined with machine-learning models trained on consented traffic patterns, to estimate the behaviour of non-consented visitors. The estimates appear only when reporting identity is set to Blended, and only for properties that meet Google's behavioural modelling thresholds: at least 1,000 events a day with analytics_storage denied for 7 days, and at least 1,000 users a day with it granted on 7 of the previous 28 days. Below those thresholds, reports show only consenting visitors, with no estimate. Google publishes no general "percentage recovered" figure for GA4, so treat any such number with caution.

Basic mode vs advanced mode

Consent Mode has two configurations, and the difference matters significantly:

Basic mode blocks all Google tags until consent is granted. No pings are sent, no modelling occurs, and the traffic drop remains fully visible in your reports.

Advanced mode loads Google tags immediately in a denied state. Cookieless pings are sent from the first page view. When a visitor later grants consent, the tags switch to full measurement. This is the configuration that enables conversion modelling and behavioural modelling.

Advanced mode sends those pings before the visitor has chosen. Whether that needs consent under Article 5(3) of the ePrivacy Directive is debated: the European Data Protection Board's Guidelines 2/2023 on the technical scope of Article 5(3) read the rule to cover tracking pixels and similar requests, and the pings carry a timestamp, user agent and referrer. Some sites choose basic mode for that reason, so weigh the data against the risk with your adviser.

Your consent tool must support Consent Mode v2 and correctly signal ad_storage, analytics_storage, ad_user_data, and ad_personalization parameters to Google tags. You can verify this with the Consent Mode checker.

Server-side tagging and cookie consent

If your site uses a server-side Google Tag Manager container, the cookie banner can cause an even more dramatic data loss than with client-side tagging alone.

Here is why. Server-side GTM depends on the client-side GTM snippet to forward data. If your cookie banner blocks the client-side GTM script entirely until consent is granted, the server container receives nothing at all, not even cookieless pings. The result is a complete measurement blackout for non-consented visitors.

Advanced consent mode is the option that changes this. It allows the client-side GTM snippet to load immediately and send cookieless pings to the server container, which forwards them to GA4, with the same Article 5(3) question as above. The server container then processes both denied-state pings and full measurement data from consented visitors.

Without Consent Mode, server-side tagging does not shrink the consent gap at all.

How to verify your Consent Mode setup

A misconfigured setup can leave you with no data recovery at all. Follow these steps to check yours:

1. Open your website in an incognito or private browser window

2. Do not interact with the cookie banner

3. Open DevTools (F12) and switch to the Network tab

4. Filter requests by collect or google-analytics

5. Look for requests going to Google. If you see them, advanced consent mode is working. The URL parameters will include gcs=G100 or similar, indicating denied consent state

6. Click Accept on the banner

7. The subsequent requests should show gcs=G111, indicating granted consent

If step 5 shows zero Google requests, you are running basic mode or the tag is fully blocked until consent. GA4 then receives nothing from non-consented visitors and has nothing to model.

Google Tag Assistant (tagassistant.google.com) also displays the consent state for each tag, which helps you confirm that denied and granted states are signalled correctly.

Use Search Console to prove traffic has not changed

When a client or stakeholder sees the GA4 drop and assumes something is wrong, Google Search Console provides the simplest proof that actual traffic is stable.

Search Console measures clicks and impressions independently of cookie consent. It tracks how often your pages appear in Google results and how often users click through, none of which requires a cookie on the visitor's device.

Compare the period before and after your banner went live. If Search Console clicks remain flat while GA4 users dropped, the banner is the likely cause. This comparison is the fastest way to show that your website is not losing visitors. It is losing measurement of visitors who declined tracking.

What you can still measure without consent

Non-consented traffic is not entirely dark. Server access logs and your hosting dashboard count requests without setting anything on the visitor's device. Cookieless analytics tools count pageviews and referrers without identifiers, although their own consent position depends on how they work and where your visitors are.

In the UK, the rules moved on 5 February 2026. The Data (Use and Access) Act 2025 added an exemption for storage used only to produce statistics that improve your own site, on conditions that include clear information and a free way to object. Whether a GA4 setup fits those conditions is a separate question, covered in the UK cookie exemptions guide.

Practical tips to maximise consent rates

While Consent Mode can add modelled data, a higher consent rate still means more accurate, first-party measurement. These adjustments can improve your acceptance rates without crossing into dark pattern territory:

• Use clear, plain language on the banner. Avoid legal jargon that makes visitors suspicious

• Apply your brand colours and logo, so the banner looks like part of your site rather than a generic white popup

• Show the banner as soon as the page loads, so visitors see the choice before they start reading

• Keep the Reject option equally accessible. EU regulators require this, and banners that hide the reject option risk enforcement action and erode trust

• Store consent in a persistent cookie so returning visitors are not prompted again

• Use region rules: show opt-in banners to EU visitors, check whether the UK statistics exemption applies to your UK analytics, and in US states that require it, show an opt-out link and honour Global Privacy Control signals

Small improvements compound. Moving from 55% to 65% consent means 10% more directly measured sessions, which in turn gives Google's modelling more consented data to learn from.

Short answers on GA4 and cookie banners

Get your Consent Mode setup right

The traffic drop after adding a cookie banner is expected, but it helps to know whether your setup sends what you think it sends. Run the Cookie scanner to see which cookies are active before consent, and check the denied and granted signals with the Consent Mode checker.

Start free