What a cookie banner on a landing page has to do. Without ruffling any feathers

Alex · Founder of Magpie AI
Published · 9 min read

In short
Under UK and EU rules, a cookie banner holds back every cookie, pixel and tag that isn't essential until the visitor says yes, offers Reject as plainly as Accept, and lets them change their mind later. Since February 2026, UK sites may count visits unasked if the figures stay theirs and visitors can opt out. On a phone, a slim bottom bar keeps your headline in view.
Contents
A cookie banner has two jobs in the UK and the EU. It holds back everything that isn't essential until the visitor agrees, and it makes saying no as easy as saying yes. The size, the place and the wording are yours to choose, and that's where most of the harm to a first screen gets done.
The first job is the one that fails quietly. The page looks fine, the banner sits there politely, and the advertising pixels have already fired behind it.
What UK and EU rules ask of a cookie banner#
Nothing that needs consent may run until the visitor gives it. The ICO's final guidance on storage and access technologies, published on 29 April 2026, says that where no exception applies "you must obtain prior consent", and that silence or a visitor who ignores the banner doesn't count as a yes. It reaches past cookies, too. The ICO lists tracking pixels, web storage, fingerprinting and "scripts and tags", so an advertising pixel that sends data without setting a cookie is still in scope.
Then comes how the choice is offered. The ICO's checklist wants a banner that "makes it as easy to refuse consent as it is to accept", and its good example puts "accept all" and "reject all" side by side and equally prominent, with any finer switches off until the visitor turns them on. Its bad example has Accept and no way to reject everything at once.
The EU's regulators say much the same. In the EDPB's cookie banner taskforce report, adopted on 17 January 2023, "a vast majority" of the data protection authorities treated a banner with no way to refuse, on any layer that has a consent button, as an infringement (a few disagreed, as the ePrivacy Directive doesn't spell out a reject option). The same report says pre-ticked boxes don't give valid consent, and nor does a Reject tucked into a paragraph as a link with nothing to draw the eye to it. France's CNIL sent formal notices over exactly that faint link in December 2024, saying "rejecting cookies should be just as easy as accepting them".
Last, a way back. The ICO says visitors must be able to withdraw consent at any time "with the same ease that they gave it", and the EDPB suggests a small icon or a link in a standard place, which for most sites means the footer.
I'm not a lawyer and this isn't legal advice. Here are the four side by side, from the regulators' own pages.
| The banner has to | ICO, UK (2026) | EDPB, EU (2023) |
|---|---|---|
| Hold back what needs consent | Prior consent; silence isn't consent | No cookie needing consent set without it |
| Offer Reject as plainly as Accept | Equally prominent accept all and reject all | No refuse option on any layer is an infringement |
| Leave the choices off until a yes | Finer switches off by default | Pre-ticked boxes aren't valid consent |
| Let visitors change their mind | Withdraw as easily as they gave it | An icon or a link in a standard place |
The ICO has been checking. In December 2025 it said 979 of the UK's 1,000 busiest websites passed its tests, 564 of them only after failing first, and in April 2026 it put the figure at 99%. Those are big sites with lawyers. When we visited sixteen software and shop sites in a headless browser from a UK address in November 2026, three set advertising cookies while their banner was still waiting for an answer, and one carried on after Reject was pressed. Sixteen is a thin sample, so treat it as a hint and check your own. If this part is wrong, no amount of banner design will put it right.
Whether your analytics need a banner#
In the UK, possibly not any more. The Data (Use and Access) Act 2025 added an exception to PECR for statistics, fully in force since 5 February 2026. Analytics can run without asking when they only count how your site is used, the figures aren't shared except with whoever helps you work them out, visitors are told what it's for, and they get a simple, free way to object.
The ICO reads that narrowly. An analytics company can do the counting for you, but only as your processor, and once the data is linked to adverts it needs consent again ("Online advertising purposes are not exempt from PECR's consent requirements and never have been"). Total visits, average scroll depth, A/B tests, load speed and bounce rates are on its list of what's likely covered, while recordings of individual visitors and what they did are not. So a Meta pixel, a LinkedIn tag, a session recorder or analytics feeding your ad campaigns still need a yes first.
If a landing page only needs to know how many people came and which section they reached, and nothing else on it needs consent, I'd set the analytics up to do just that and drop the banner for UK visitors. You'd still need a line in your privacy policy (one that matches what your site promises) saying what's counted and why, and an opt-out a visitor can actually find, such as a footer link to a page with a switch. The ICO doesn't fix the form, though it says browser settings alone won't do.
Elsewhere in Europe it depends on the regulator. France's CNIL has let audience measurement run unasked since 2020, on its own conditions: one publisher only, the last part of the IP address dropped, trackers kept 13 months at most, and visitors told and able to object. It also says "Most large audience measurement offerings do not fall within the scope of the exemption, regardless of their configuration", which is worth reading twice before you decide your setup qualifies.
What a cookie banner costs your first screen#
On a phone it can take a big share of it. Google's web.dev guide says a cookie notice "typically takes up a larger portion of the screen" on mobile, sometimes enough that its own text becomes the largest thing the page draws, which is the element your load time is measured by.
Here's a made-up seed subscription shop at sproutbox.example, on a 393 by 852 pixel phone screen (an iPhone 14 Pro's).
| Before | After | |
|---|---|---|
| Where it sits | A card in the middle of the screen | A bar along the bottom |
| Height | 380px, 45% of the screen | 150px, 18% of the screen |
| Words | Three paragraphs about partners and purposes | One sentence and a link to the full policy |
| Buttons | Accept all, Manage options | Accept and Reject, the same size and style |
| What it covers | The line under the headline and the button | The bottom 150px of the screen only |
The before version also breaks the rules above, since Reject sits a layer down. The after version leaves the headline, the line under it and the Start my box button where the first screen needs them.
What it costs in sign-ups, I can't tell you. I haven't found a study that measures it on landing pages. The web.dev guide warns that "smaller sites may find that users bounce rather than accept the cookie notice of an unfamiliar site with obscured content", which is Google's advice, not a measurement.
What has been measured is how a banner's design steers the answer. Nouwens and colleagues (CHI 2020) scraped the five most popular consent tools on the UK's top 10,000 sites and found only 11.8% of the 680 banners met the minimal legal requirements they set. In their experiment with 40 people, taking Reject off the first page raised consent by 22 to 23 percentage points. That's the gain a hidden Reject buys, and it's the design the ICO and the EDPB call out. Utz and colleagues (CCS 2019) tested notices on a German site with more than 80,000 visitors and found people were more likely to deal with one shown in the lower part of the screen, the left especially. That's handy for a landing page, because the bottom of a phone screen is the part your headline doesn't need, and a banner that gets answered goes away.
A banner can also make a page feel slow. Web.dev calls cookie notices "a very common source of layout shifts" when one is slotted in at the top and pushes everything down, and pressing Accept can load a pile of other sites' scripts at once and stall the next tap. A visitor reaching for your button as the page lurches is a sitting duck. The fixes are the same ones in how fast a landing page should load on a phone.
A cookie banner that does its job and leaves the headline alone#
This is what I'd put on a landing page, from the rules and the guidance above.
- A bar along the bottom, drawn over the page so nothing moves when it appears. Eyes go to the top left first (where visitors look first), so that spot stays your headline's.
- One plain sentence on what you'd like to use and why, with a link to the cookie policy for the detail.
- Accept and Reject with the same size, colour and weight, and a quieter Settings link beside them if you offer finer choices.
- Advertising, recording and anything else that needs a yes loaded only after Accept, from the banner's own answer, never from the page's load.
- The banner's script loaded asynchronously and straight from your page rather than through a tag manager, as web.dev suggests, and a Cookie settings link in your footer for changing your mind.
I wouldn't agonise over the wording, as long as both buttons say what they do. "Accept" and "Reject" are fine. A bright Accept beside a faint "continue without accepting" link is the pattern the EDPB and the CNIL called out.
How to check what your site runs before anyone answers#
Open your site in a private window from a UK or EU connection, and before you touch the banner, open the browser's developer tools. Look at the cookies the page has set and at the network requests it's made, searching for names such as facebook, doubleclick, linkedin, clarity or hotjar. Anything advertising or recording that's already there ran before consent. Then press Reject, reload and look again.
Where you check from matters, because some banner tools show visitors outside the UK and EU no banner at all and load everything. A check from New York can make a careful site look careless.
Magpie's full report does this in code on its phone visit. A tracker counts as running only when it sets its cookie or sends data (a script that loads and waits doesn't), and what's seen is split at the moment the banner is answered, so the report shows what ran before anyone said yes and what still ran after Reject. Consent is judged only when the visit comes from the UK, the EU or the EEA and the banner was seen, and the report also checks that your privacy policy names each tracker's company. For now our visits come from the US, so the report lists what runs and doesn't yet judge consent.
Questions people ask#
Will a full-screen cookie banner hurt my Google ranking?#
Not by itself. Web.dev says "Google Search does not penalize the usage of interstitials when they are used to comply with legal regulations such as in the case of cookie banners". Visitors are another matter, and few of them are happy as larks about a screen they have to clear before they can read anything.
Do I need a banner if my site sets no cookies?#
It depends on what else the page does, since the rules cover pixels, local storage, fingerprinting and scripts as well as cookies. If the only things stored are strictly necessary (the ICO's examples include a shopping basket, a logged-in session and security for a service someone asked for), you need no consent and no banner. If your analytics fit the UK's statistics exception, you need the explanation and the opt-out but not the question. The moment an advertising pixel or a session recorder goes on, you're back to asking first, and "we don't use cookies" stops being the test.
Sources
- ICO, Guidance on the use of storage and access technologies (final, 29 April 2026) (opens in a new tab)
- ICO, How do we manage consent in practice? (opens in a new tab)
- ICO, What are the exceptions? (opens in a new tab)
- ICO, What are storage and access technologies? (opens in a new tab)
- ICO, ICO action secures increased cookie compliance (4 December 2025) (opens in a new tab)
- ICO, Final storage and access technologies guidance published (29 April 2026) (opens in a new tab)
- Data (Use and Access) Act 2025, section 112 (opens in a new tab)
- Data (Use and Access) Act 2025, Schedule 12 (new PECR Schedule A1) (opens in a new tab)
- The Data (Use and Access) Act 2025 (Commencement No. 6 and Transitional and Saving Provisions) Regulations 2026 (opens in a new tab)
- EDPB, Report of the work undertaken by the Cookie Banner Taskforce (January 2023) (opens in a new tab)
- CNIL, Dark patterns in cookie banners: CNIL issues formal notice to website publishers (December 2024) (opens in a new tab)
- CNIL, Sheet n°16: Use analytics on your websites and applications (2020) (opens in a new tab)
- Nouwens, Liccardi, Veale, Karger and Kagal, Dark Patterns after the GDPR (CHI 2020) (opens in a new tab)
- Utz, Degeling, Fahl, Schaub and Holz, (Un)informed Consent: Studying GDPR Consent Notices in the Field (CCS 2019) (opens in a new tab)
- web.dev, Best practices for cookie notices (Katie Hempenius and Barry Pollard) (opens in a new tab)

About the author
Alex · Founder of Magpie AI
Alex founded Magpie AI and writes about what turns a landing page's visitors into customers, drawing on what its audits find on real sites, from the first screen to the checkout


