Cookie consent banners: what the evidence shows — ROR Labs cover showing 11.8% lawful.
|

Cookie Consent Banners: What the Evidence Shows

Key takeaways · 11 min read

  • European law requires consent for non-essential cookies, and refusing must be as easy as accepting.
  • Only 11.8% of 680 UK sites met minimal legal requirements in one study; 89% of banners in another violated the law.
  • Removing the reject button from the first page raised consent by about 22 points; banner design drives most choices.
  • Some sites recorded consent before users chose, or after they opted out.

Almost every website in Europe now greets visitors with the same small negotiation. A box slides up: this site uses cookies. A large bright button says Accept all, and somewhere, sometimes, a smaller grey link says Manage options. Most people click the bright button without reading a word, dozens of times a week, and few ever wonder what they have agreed to.

Those banners exist because of European law. Sites have needed consent before storing most non-essential cookies on a device since 2009, and since 2018 the General Data Protection Regulation has set a strict standard for what consent means: freely given, specific, informed and unambiguous. The idea was to give people real control over tracking.

Researchers have now measured what the banners do. They have scraped thousands of them, tested them on tens of thousands of real visitors, and checked whether sites respect the choices people make. Small design decisions change consent rates by tens of percentage points, most banners do not meet the legal standard, and some sites record consent that was never given. This article looks at that evidence and at what regulators have done about it.

A cookie is a small text file that a website stores in your browser. Some are essential: they keep you logged in or remember what is in your basket. Others exist mainly to recognise you again later, on the same site or across many sites, so that advertisers can build a profile of what you read and buy. The main distinction is between first-party cookies, set by the site you are visiting, and third-party cookies, set by other companies whose code is embedded in the page.

A typical news or shopping site can load code from dozens of such companies, so clicking accept often means agreeing on behalf of all of them at once, a list that may run to hundreds of named vendors hidden behind a settings link. The banner asks about cookies, but the same companies can also use other techniques, such as device fingerprinting, that do not rely on cookies at all and that a banner may not mention.

What the law requires

Two pieces of European law apply. The ePrivacy Directive, in its Article 5(3) as amended in 2009, requires consent before a site stores or reads information on a user’s device, unless it is strictly necessary for a service the user asked for. The GDPR defines what valid consent is, and its Article 7(3) adds that it must be as easy to withdraw consent as to give it.

The EU’s top court has filled in the detail: in the Planet49 case in 2019, the Court of Justice ruled that a pre-ticked checkbox is not valid consent. Since then regulators have said that refusing should be as easy as accepting, that consent cannot be assumed from simply continuing to browse, and that the purposes of tracking must be stated clearly.

What researchers found when they looked

Midas Nouwens and colleagues presented a study at the CHI conference in 2020 in which they scraped the consent pop-ups of 680 UK websites among the top 10,000 that used the five most popular consent management platforms, the third-party services that many websites install to handle consent. Only 11.8% met a minimal set of requirements based on European law: explicit consent, rejecting as easy as accepting, and no pre-ticked boxes.

Only 12.6% of sites showed a ‘reject all’ button on the same page as ‘accept all’, which was the most common failure. For everyone else, saying no meant clicking through to another screen, sometimes several, and switching off purposes or vendors one by one.

Consent pop-ups on 680 UK websites

Sites using the five most popular consent management platforms, 2019.

11.8%met minimal requirements based on European law
12.6%offered ‘reject all’ on the first page

Nouwens M, Liccardi I, Veale M, Karger D, Kagal L, Proceedings of CHI 2020, 1–13.

A pointillist illustration: a large laptop screen on a desk at night showing a web page half covered by a pop-up with one bright amber button and a small grey link beside it.
A consent pop-up with one bright button. Design, not preference, drives most choices.

Design changes the answer

The same team then ran a field experiment with 40 participants, showing them different versions of the pop-up on sites they actually visited. Removing the ‘reject all’ button from the first page raised the probability of consent by about 22 percentage points. Putting more detailed choices about purposes or vendors on the first page lowered consent by between 8 and 20 points. The answer was the same whether the notice was a banner or a barrier that blocked the page, but banners were ignored 3.6 times as often.

Larger experiments confirm this. In 2019, Christine Utz and colleagues tested consent notices on more than 80,000 visitors to a German website. People were more likely to interact with notices in the lower left of the screen. More people accepted tracking when given a simple yes-or-no choice than when they had to allow each category or company individually, and very few chose specific categories unless they were pre-selected. Nudges such as highlighted buttons and pre-ticked boxes had large effects.

In a randomised field trial with 1,493 users, published in 2021 by Jan Michael Bauer, Regitze Bergstrøm and Rune Foss-Madsen, manipulating the banner’s design increased consent by 17 percentage points of the whole sample. The authors concluded that the ability of site owners to steer choices in this way is probably at odds with what the law intends.

Field tests of banner design

Experiments on real website visitors.

+17 ptsconsent gained by manipulating banner design, 1,493 users
80,000+visitors in the largest field test of consent notices

Bauer JM, Bergstrøm R, Foss-Madsen R, Computers in Human Behavior, 2021; Utz C et al, ACM CCS 2019.

How design changes consent

Effect of pop-up design on the probability of accepting, field experiment.

Design changeEffect on consent
Remove ‘reject all’ from first pageUp about 22 points
Show vendor choices on first pageDown about 20 points
Show purpose choices on first pageDown about 9 points
Banner instead of blocking barrierNo change; ignored 3.6× as often

Nouwens M et al, CHI 2020. Small sample: 40 participants.

Do sites respect the choice?

Behind the banner, the consent decision is stored and passed on to advertising companies. In 2020, Célestin Matte, Nataliia Bielova and Cristiana Santos examined sites using the Transparency and Consent Framework, an industry standard run by the advertising trade body IAB Europe that many consent platforms follow.

A pointillist illustration: a small round cookie on a saucer beside a keyboard, with a tiny amber tag attached by a thread.
A cookie on a saucer. What is stored behind the banner matters more than its buttons.

Among 1,426 websites with such banners, 141 registered positive consent before the user had made any choice, 236 nudged users by pre-selecting options, and 27 stored positive consent even after the user had explicitly opted out. In more detailed tests of 560 sites, at least one suspected violation appeared on 54%. The authors built a browser extension to help regulators and users detect such problems.

Banner wording has problems too: Cristiana Santos and colleagues annotated around 400 banners on popular English-language websites visited from the EU and found that 89% violated applicable law in what they said. Vague purposes such as ‘improving user experience’ were the most common problem, in 61%. In 30%, positive framing undermined the requirement that consent be freely given and informed.

What happens behind the banner

Sites using the IAB Europe Transparency and Consent Framework.

FindingSites
Positive consent registered before any choice141
Options pre-selected to nudge acceptance236
Positive consent stored after opting out27
At least one suspected violation, of 560 tested54%

Matte C, Bielova N, Santos C, 2020 IEEE Symposium on Security and Privacy, 791–809. 1,426 sites with such banners were found among 28,257 crawled.

What regulators have done

In January 2022, France’s data protection authority, the CNIL, fined Google a total of €150 million and Facebook’s parent company €60 million because their sites made refusing cookies harder than accepting them: accepting took one click, refusing took several. The companies were ordered to offer an equally simple way to refuse.

The consent framework itself was challenged in Belgium. The Belgian data protection authority ruled in 2022 that IAB Europe’s framework infringed the GDPR, and in March 2024 the Court of Justice of the European Union, in case C-604/22, held that the coded string recording a user’s preferences can be personal data and that the organisation running such a framework can share responsibility for how it is used.

Regulators have also moved in the other direction on low-risk cookies. The United Kingdom’s Data (Use and Access) Act 2025 created exemptions from the consent requirement for some cookies considered low risk, such as certain analytics, reflecting a wider view that banners for everything had produced fatigue and little control.

What banners say

About 400 banners on popular English-language websites, assessed against EU law.

89%of banners violated applicable law in their wording
61%described their purposes too vaguely to be valid

Santos C, Rossi A, Sánchez Chamorro L, and colleagues, Proceedings of the 20th Workshop on Privacy in the Electronic Society, 2021, 187–194.

Outside Europe

The European approach is based on opting in: nothing non-essential should happen until you agree. The United States has mostly taken the opposite route. California’s privacy law, as amended by the California Privacy Rights Act, gives residents the right to opt out of the sale or sharing of their personal information, and the state’s regulations require businesses to treat a browser-level opt-out signal, such as Global Privacy Control, as a valid request. Several other US states have since passed similar laws.

In practice the rules form a patchwork. Many international websites show a full consent banner only to visitors from Europe or the UK, a simpler notice to Californians and nothing at all to people elsewhere. The same site can therefore track two readers of the same article very differently depending on where they appear to be. For readers outside these jurisdictions, browser settings are often the only control available.

A pointillist illustration: a smartphone held in a hand on a train, its screen showing a pop-up with two buttons of equal size, one blue and one amber.
A pop-up with two equal buttons. When refusing is as easy as accepting, many more people refuse.

Are banners the right tool at all?

Critics on all sides agree that the banners have not worked as intended. People face so many that clicking accept has become a reflex. Utz and colleagues described ‘consent fatigue’ in 2019, and the finding that banners are ignored far more than barriers suggests many people simply work around them.

The research also shows that when refusing is made as easy as accepting, many more people refuse. That suggests the banners largely measure design, and that a fair design would reveal a lower appetite for tracking than current consent rates imply. Some researchers and regulators have argued for moving the choice to the browser, so that a single setting such as the Global Privacy Control signal applies to every site, rather than asking on each page.

Publishers have a lot at stake, which explains why the fight over design is so intense. Many free websites earn most of their revenue from advertising, and personalised adverts usually sell for more than untargeted ones. Every percentage point of consent therefore has a price, which gives sites a strong incentive to make accepting easy and refusing tedious. The research suggests that the design of the banner often decides the outcome.

What you can do

Individuals have a few modest but real options. Look for a reject or ‘necessary only’ option before accepting; on sites that follow the rules, it should be on the first screen. Browser settings that block third-party cookies, and extensions that answer banners automatically with a refusal, reduce the number of choices you have to make. Clearing cookies from time to time removes identifiers that have built up, and private browsing windows discard them when you close the window.

Cookies are only one tracking method among many. We looked at how location data from phones is collected and sold in our article on data brokers, and at the wider tricks of manipulative design in our article on dark patterns.

This section collects what the UK ICO, European regulators and California agencies say about cookie consent banners, and what is still only a proposal. It is general information, not legal advice.

Source and what it says
SourceWhat it says
UK ICO, How do we manage consent in practice?Refusing must be as easy as accepting: its good-practice examples show “accept all” and “reject all” with equal prominence. Pre-ticked boxes and implied consent by continued use are poor practice. The page says it is under review after the Data (Use and Access) Act became law on 19 June 2025.
UK ICO, cookie compliance update (Dec. 4, 2025)979 of the UK’s top 1,000 websites met its checks at the latest test and 21 still failed. It tested whether advertising cookies were stored before a choice, whether rejecting was as easy as accepting, and whether cookies were set without consent.
GOV.UK, Data (Use and Access) Act factsheet (June 27, 2025)Adds a new exception for cookies that collect usage statistics to improve a service, without consent. Commencement is staged and the factsheet gives no cookie date.
CNIL (France), misleading banners (Dec. 12, 2024) and the EDPB taskforce report (Jan. 2023)CNIL gave publishers one month to change banners where the refuse link was styled less prominently or blended into text. The EDPB taskforce says most authorities treat a banner with no refusal option at the same level as accept as a breach, and judge other designs case by case.
California Privacy Protection Agency, Enforcement Advisory 2024-02 and Attorney General, CCPA pageThe advisory says agreement obtained through dark patterns is not consent, and that “Accept All” and “Decline All” are symmetric. The Attorney General says businesses must honour opt-outs, including Global Privacy Control signals, within 15 business days.
European Parliament, procedure 2025/0360(COD)The Commission’s Digital Omnibus proposal of Nov. 19, 2025 would move cookie rules into the GDPR. The file shows it at committee stage, with amendments tabled July 27, 2026.

What we could not confirm: nothing we read shows the EU proposal has been adopted, so the existing ePrivacy rules still apply as before. The ICO’s own pages disagree: one says there is no statistics exemption, another lists one under the new Act, and no page gives a commencement date, so we cannot say which is in force. The ICO’s April 2026 final guidance and a Council press release of Oct. 21, 2026 were blocked, so we report nothing from them. We read no official UK or EU page telling users how to refuse cookies in a browser, and none that requires sites to honour Global Privacy Control. California frames this as an opt-out system, and the UK and EU as opt-in.

What to do

Steps sorted by when they matter. Each one says how strong the evidence is: Measured Official advice Untested

Before

  • Block third-party cookies in your browser settings, or use an extension that answers banners with a refusal. Untested see Sources
  • Clear cookies from time to time, and use private windows, which discard them when closed. Untested see Sources

If it happens

  • Look for “reject” or “necessary only” before accepting. On sites that follow the rules it should be on the first screen; in one study only 12.6% put “reject all” there. Measured see Sources

What not to do

  • Do not assume refusing cookies stops all tracking. Techniques such as device fingerprinting do not rely on cookies at all. Untested see Sources

Questions people ask

Do I have to accept cookies?

No. Under EU law, non-essential cookies need your consent, and refusing should be as easy as accepting. Strictly necessary cookies do not need consent.

How many cookie banners follow the law?

Few. One study found only 11.8% of 680 UK sites met minimal requirements; another found 89% of about 400 banners violated the law in their wording.

Does rejecting cookies actually work?

Usually, but not always. One study found 27 sites that stored positive consent even after users opted out.

Are cookie walls allowed?

Regulators have generally said that access to a site should not depend on accepting all tracking, though some accept a paid alternative. The rules and their interpretation are still being argued over.

Why is the accept button always bigger?

Because design changes behaviour. Removing the reject button from the first page raised consent by about 22 percentage points in one experiment.

Have regulators fined companies over cookie banners?

Yes. In 2022 France fined Google €150 million and Facebook’s parent €60 million for making refusal harder than acceptance.

The short version

  • European law requires consent for non-essential cookies, and refusing must be as easy as accepting.
  • Only 11.8% of 680 UK sites met minimal legal requirements in one study; 89% of banners in another violated the law.
  • Removing the reject button from the first page raised consent by about 22 points; banner design drives most choices.
  • Some sites recorded consent before users chose, or after they opted out.
  • Regulators have fined Google and Facebook, and the EU court has tightened rules on consent frameworks.

This article summarises published research and regulatory decisions about cookie consent. It is not legal advice for website owners or users. Rules differ between the EU, the UK and other countries and continue to change.

Further reading. Nouwens and colleagues, ‘Dark Patterns after the GDPR: Scraping Consent Pop-ups and Demonstrating their Influence’, CHI 2020, is open access and readable. Matte, Bielova and Santos, IEEE Symposium on Security and Privacy, 2020, looks behind the banner.

Sources

  1. Nouwens M, Liccardi I, Veale M, Karger D, Kagal L. Dark patterns after the GDPR: scraping consent pop-ups and demonstrating their influence. Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, 1–13. doi:10.1145/3313831.3376321.
  2. Utz C, Degeling M, Fahl S, Schaub F, Holz T. (Un)informed consent: studying GDPR consent notices in the field. Proceedings of the 2019 ACM SIGSAC Conference on Computer and Communications Security, 973–990. doi:10.1145/3319535.3354212.
  3. Bauer JM, Bergstrøm R, Foss-Madsen R. Are you sure, you want a cookie? The effects of choice architecture on users’ decisions about sharing private online data. Computers in Human Behavior, 2021;120:106729. doi:10.1016/j.chb.2021.106729.
  4. Matte C, Bielova N, Santos C. Do cookie banners respect my choice? Measuring legal compliance of banners from IAB Europe’s Transparency and Consent Framework. 2020 IEEE Symposium on Security and Privacy, 791–809. doi:10.1109/SP40000.2020.00076.
  5. Santos C, Rossi A, Sánchez Chamorro L, and colleagues. Cookie banners, what’s the purpose? Analyzing cookie banner text through a legal lens. Proceedings of the 20th Workshop on Privacy in the Electronic Society, 2021, 187–194. doi:10.1145/3463676.3485611.
  6. Directive 2002/58/EC (ePrivacy Directive), Article 5(3), as amended by Directive 2009/136/EC.
  7. Regulation (EU) 2016/679 (General Data Protection Regulation), Articles 4(11) and 7(3).
  8. Court of Justice of the European Union. Case C-673/17, Planet49, 1 October 2019; Case C-604/22, IAB Europe, 7 March 2024.
  9. Commission nationale de l’informatique et des libertés (CNIL). Deliberations on Google LLC, Google Ireland Limited and Facebook Ireland Limited, 31 December 2021, announced January 2022.
  10. ICO: consent in practice, compliance update. GOV.UK factsheet; CNIL; EDPB report; CPPA; California AG.

Similar Posts