
Pre-ticked boxes and GDPR: cookie options already on
At a glance
EU-21HighePrivacy Art. 5(3); GDPR Art. 4(11) and recital 32; CJEU C-673/17 Planet49 (2019) and C-61/19 Orange România (2020); EDPB Cookie Banner Taskforce report (2023), paras. 9 and 10
Peeky opens your cookie banner's settings and reads every switch before touching one. Then it saves them untouched to see whether that's stored as a yes.
Last checked against the source:
The rule
EU law doesn’t mention toggles, sliders or settings screens. It says what consent is, and a choice someone else made in advance isn’t it.
Article 4(11) of the GDPR defines consent as a “freely given, specific, informed and unambiguous indication of the data subject’s wishes by which he or she, by a statement or by a clear affirmative action, signifies agreement”. Recital 32 adds that ticking a box can be consent, then says what can’t: “Silence, pre-ticked boxes or inactivity should not therefore constitute consent.” Article 5(3) of the ePrivacy Directive, which covers storing or reading information on a visitor’s device, uses the same idea of consent.
On 1 October 2019 the Court of Justice ruled on an online lottery whose sign-up form carried a pre-ticked box agreeing to advertising cookies. Consent “is not validly constituted if, in the form of cookies, the storage of information or access to information already stored in a website user’s terminal equipment is permitted by way of a pre-checked checkbox which the user must deselect to refuse his or her consent.” A preselected tick “does not imply active behaviour on the part of a website user” (para. 52). The answer doesn’t change whether or not the information stored is personal data (operative part, point 2).
A year later the Court looked at paper mobile phone contracts where the company had ticked a consent box before the customer signed. That tick proved nothing: “the mere fact that that box was ticked is not such as to establish a positive indication” of consent (para. 46). The company has to show that its customers gave consent “by active behaviour”, “with the result that that company cannot require them actively to express their refusal” (para. 51). The same burden falls on a website that asks for cookie consent.
The EDPB’s consent guidelines say it flatly: “The use of pre-ticked opt-in boxes is invalid under the GDPR” (Guidelines 05/2020, para. 79), and controllers may not offer “pre-ticked boxes or opt-out constructions that require an intervention from the data subject to prevent agreement” (para. 81). The authorities’ cookie taskforce then looked at the exact screen this check reads.
The taskforce found that “several controllers provide users with several options (typically, representing each category of cookies the controller wishes to store) with pre-ticked boxes on the second layer of the cookie banner (after the user clicked on the ‘Settings’ button of the first layer).” Its members “confirmed that pre-ticked boxes to opt-in do not lead to valid consent as referred to either in the GDPR (see in particular recital 32 …) or in Article 5(3) of the ePrivacy Directive.”
National guidance goes into the detail of the settings screen. The CNIL’s recommendation asks for checkboxes “décochées par défaut” (unticked by default) and sliders “désactivés par défaut” (switched off by default). The Spanish AEPD’s guide, updated in May 2024, covers the Save button: if a visitor saves without selecting any cookie, that “equivaldrá al rechazo de todas las cookies” (counts as rejecting all cookies), and pre-marked options in favour of accepting are never acceptable. The Italian Garante’s guidelines of 10 June 2021 say a visitor who keeps the default settings hasn’t consented, and that choice has to be recorded. Germany’s DSK says the options “nicht ‘aktiv’ voreingestellt sein dürfen” (must not be preset to active, OH Digitale Dienste, para. 118). The Irish DPC is the most direct: “You are not permitted to use pre-checked boxes, sliders or other tools set to ‘ON’ by default to signal a user’s consent to the setting or use of cookies.”
The industry’s own consent framework says the same. IAB Europe’s Transparency and Consent Framework requires that in the layer with a choice per purpose “the default choice must be ‘no consent’, ‘no opt-in’ or ‘off’”, and its technical specification says that while the banner shows, every consent signal is set to false.
Cookie rules are enforced under each country’s own ePrivacy law, and the consent conditions also sit in the GDPR’s upper fine tier (Art. 83(5)(a)). In the decisions we found, pre-ticked options ended in orders, reprimands or fines between €4,000 and €50,000, usually alongside other cookie findings. The Kruidvat fine started higher and was cut on objection.
What PeekWell checks and how
EU-21 asks one question: is any non-essential cookie option switched on before the visitor chooses? It’s pass or fail.
The scan is automated. It opens your public pages the way a first-time visitor’s browser would, opens the banner’s settings layer from the first screen (Settings, Customise, Manage options or whatever it’s called) and reads every switch and box before touching any of them. That includes vendor lists the layer links to. It takes a screenshot of the layer as it opened. The check fails on any of three things:
- A non-essential option starts on. Analytics, statistics, performance, advertising, marketing, personalisation, social media and session recording all count, and so does “functional” when it covers third-party services. A strictly necessary category shown as on and locked is expected and doesn’t count.
- Saving the untouched layer does more than refusing. The scan presses Save or Confirm my choices without changing anything, reloads, visits two more pages, and compares what it sees with the run where it pressed Reject all. Any extra non-essential cookie, storage or request fails, and so does consent stored for a non-essential purpose or vendor.
- The first screen only offers a pre-selected bundle. If, besides accepting, the first screen’s only control is a button that saves a pre-selected set (“Save”, “Confirm my choices”, “OK”) and that set includes non-essential purposes, the check fails.
On pages that run the IAB framework, the scan also reads the framework’s consent data before anything is chosen. Consent recorded at that point for a non-essential purpose or vendor counts as a switch that starts on. A switch can look off and still let a tag through, which is why the scan compares the switches with what happens after Save.
The check passes when every non-essential option starts off and saving the untouched layer gives the same result as Reject all. A banner with no settings layer has nothing pre-set, so it can’t fail the first two. Where the line falls is PeekWell’s own reading of the rules above; the finding shows the screenshot and the before and after.
One case stays open. In France the CNIL treats audience measurement limited to the publisher’s own statistics as strictly necessary, so it needs no consent (guidelines of 17 September 2020, paras. 50 and 51), and the Garante allows something similar for analytics cookies in Italy under conditions. A site there might show such a tool as a switch that’s already on. EU-21 still reports that switch, because from outside the scan can’t confirm the tool stays inside those limits.
Switches labelled legitimate interest, or “object”, aren’t part of this check. They belong to EU-22, which looks at what stays on after Reject all. Whether the first screen makes saying no as easy as saying yes is EU-03, and trackers that start before any choice are EU-01.
With no banner at all, or a banner with no way to refuse, the answer depends on what the site does. If anything non-essential is stored or sent on any run, the check fails at the same severity. If nothing non-essential shows up at all, it passes, and the finding says which runs showed that.
The scan has limits, and the report says so. It reads the settings layer of the public banner as a first-time visitor sees it. Pre-selected options inside an app, in a sign-up flow behind a login you didn’t share, or on a newsletter form are outside this check. It doesn’t port-scan, try to get in where it isn’t allowed, run any security test or open addresses nobody linked to. A Passed means the expected behaviour was observed on the pages scanned. It doesn’t say the site complies.
Why it matters for a company
A pre-ticked setting is easy for anyone to see, and the Dutch Kruidvat decision shows what an authority does with it. The AP’s investigators walked through the cookie flow on kruidvat.nl. In the third step a slider was set to advertising cookies by default, and a visitor who went all the way through without changing anything was recorded as agreeing. On 4 September 2020 the advertising cookies were still ticked by default; by 5 October 2020 the pre-filled boxes were gone. The AP fined A.S. Watson €600,000 in a decision dated 2 May 2024, which also covered cookies set before any consent. On objection, on 27 May 2025, it set the fine at €50,000, pointing to the long procedure, the company’s full admission, the limited seriousness and a comparable cookie case. Its finding stayed: personal data had been processed without a legal basis.
The Spanish Freshly Cosmetics file shows the other half of the problem. Behind Personalizar every cookie group started as accepted, and when a visitor unticked them the site kept using cookies that weren’t technical or necessary. The proposed fine was €5,000, and the company paid €4,000 with the reduction for voluntary payment.
Under Orange România and Article 7(1) of the GDPR, it’s the company that has to show the visitor acted, and a default shows nothing. A switch that looks off can still let tags through, as the Irish DPC found in 2019 when one site’s tool kept setting marketing cookies after every box was unticked. And the fix is usually a setting in the consent tool, which makes this one of the quicker findings to close.
Smaller companies and larger companies
The rule has no size threshold. What differs is where the pre-set switches come from.
In a smaller company the banner usually comes from a plugin or a hosted builder, and the defaults were set once, by the tool or by whoever installed it. The Italian Garante’s 2024 decisions on Onlinestore and Sotto il cavolo are typical: an advertising and profiling option pre-selected under Personalizza, or six pre-ticked boxes behind a single Save and close. Both ended without a fine. The Belgian legal news site in decision 12/2019, with 35,000 monthly readers by its own count, had moved from opt-in to opt-out on purpose, and the Litigation Chamber weighed that choice when it set a €15,000 fine.
In a larger company the settings layer tends to be longer and owned by several teams. Vendor lists run to hundreds of names, and one default can switch on all of them: on the Knack and Le Vif news sites, the Belgian authority found 449 partners given consent through pre-ticked boxes. Multi-step flows add another risk, because a Save at the end of a long path is where a default turns into a recorded yes, as it did on kruidvat.nl.
Enforcement cases
Published decisions about other companies, listed for context. Each links to the authority's own page. They say nothing about any particular website.

Larger companies
A.S. Watson Health & Beauty Continental Europe B.V. (Kruidvat.nl)
The AP found that in the third step of the cookie flow on kruidvat.nl a slider was set to advertising cookies by default, and that a visitor who went through the whole flow without changing a setting was recorded as consenting. Cookies and a Google Analytics beacon were also set before any consent. On objection the AP lowered the fine and kept its view that personal data had been processed without a legal basis.
Read the Autoriteit Persoonsgegevens (Netherlands) publication about A.S. Watson Health & Beauty Continental Europe B.V. (Kruidvat.nl)Coolblue
The AP found that in 2020 the webshop's cookie statement said it assumed visitors agreed, and the boxes for consent to cookies were pre-ticked. Coolblue had changed its practice by June 2020.
Read the Autoriteit Persoonsgegevens (Netherlands) publication about CoolblueRoularta Media Group (Knack and Le Vif websites)
The inspection found that on the knack.be and levif.be news sites 449 partners or vendors were given consent by default through pre-ticked boxes. The Litigation Chamber found that the pre-ticked boxes gave no valid consent. The same decision covers other cookie findings, and the fine is for all of them.
Read the APD / GBA Litigation Chamber (Belgium) publication about Roularta Media Group (Knack and Le Vif websites)
Smaller companies
Freshly Cosmetics, S.L.
The banner's first layer offered Personalizar and Aceptar todas. Behind Personalizar every cookie group was pre-set to 'aceptados', and the AEPD found that a visitor who unticked the groups still got cookies that were not technical or necessary. By the AEPD's later check the groups started as 'no aceptados'.
Read the AEPD (Spain) publication about Freshly Cosmetics, S.L.Operator of a Belgian legal news website (name withheld in the published decision)
The inspection found pre-ticked boxes for cookie preferences on 29 May 2019. The company said it had chosen to move from an opt-in system to an opt-out one, with every box ticked to accept cookies. The Litigation Chamber found that this gave no valid consent, citing recital 32 and Planet49, and found legitimate interest wrongly relied on for statistics cookies. The fine also covers gaps in the privacy notice.
Read the APD / GBA Litigation Chamber (Belgium) publication about Operator of a Belgian legal news website (name withheld in the published decision)advocado GmbH
The Federation of German Consumer Organisations (vzbv) sued the lawyer-search platform. The court found that the banner's four boxes came pre-selected with an OK button to accept them, and that the redesigned banner still had every cookie pre-selected and switched on by a green 'Cookies zulassen' button, with deselection only after opening the details. It held that neither version gave valid consent. advocado appealed to the Oberlandesgericht Rostock; we found no published outcome.
Read the Landgericht Rostock (Germany) publication about advocado GmbHOnlinestore S.r.l.
The first layer offered Accetta, Personalizza and Altre info, with no X and no reject button. Under Personalizza two options came pre-selected: necessary cookies, which could not be deselected, and advertising and profiling cookies, which could.
Read the Garante (Italy) publication about Onlinestore S.r.l.Sotto il cavolo s.n.c.
The banner offered only Salva e chiudi and a cookie policy link, with no X to close it. Six boxes came ticked with a green mark: AddThis, Facebook, Google Analytics, Google Maps, YouTube and Zopim Chat. The company reconfigured the banner during the proceedings.
Read the Garante (Italy) publication about Sotto il cavolo s.n.c.
How to fix it
The aim is a settings layer where everything optional starts off and an untouched Save means no.
- Look at the settings as a visitor. Open your site in a private window, open the banner’s settings and write down every switch or box that’s already on, including in any vendor list.
- Set every non-essential category and vendor to off by default. Most consent tools have a default-state setting per category. Leave only the strictly necessary category on, and lock it.
- Make an untouched Save store a refusal. Press Save without changing anything, then check the browser’s network tab and the cookies. Nothing should load that doesn’t also load after Reject all.
- Check the vendor list and the framework settings. If you run the IAB framework, make sure no purpose or vendor starts with consent. Legitimate-interest switches are covered by EU-22.
- Don’t offer a pre-filled bundle on the first screen. If the first screen has Accept all, put Reject all beside it rather than a Save or OK that stores whatever is pre-ticked.
- Check your tag manager’s starting state. If it keeps its own consent defaults, they should start as denied too, so a tag can’t fire before the banner tells it to.
- Re-scan. Clear cookies and storage, reload and run a scan again. The finding should show every non-essential switch off and an untouched Save matching Reject all.
Whether a given banner meets the rule in a given country is a question for your legal adviser. Peeky reports the switches it saw and what Save did.
Questions
Are pre-ticked boxes allowed under the GDPR?
Not for consent. Recital 32 of the GDPR says that "silence, pre-ticked boxes or inactivity" shouldn't count as consent, and the EDPB calls pre-ticked opt-in boxes invalid. For cookies the same goes for a toggle or slider that starts on.
A box the visitor ticks themselves is fine. So is a strictly necessary category shown as on and locked, because it doesn't need consent in the first place.
What did the Court of Justice decide in Planet49?
In October 2019 the Court of Justice ruled on an online lottery whose sign-up form had a pre-ticked box agreeing to advertising cookies. It held that a box the user has to untick to refuse doesn't give valid consent to cookies.
It also said the answer is the same whether or not the cookie holds personal data, and that clicking to join the lottery wasn't consent to the cookies either.
Does a toggle that starts on count as a pre-ticked box?
Yes. The Irish Data Protection Commission says you may not use "pre-checked boxes, sliders or other tools set to 'ON' by default" for cookie consent, and the CNIL asks for sliders switched off by default. In the Kruidvat decision the Dutch authority treated a slider set to advertising cookies as pre-ticked boxes.
What should Save my preferences do if the visitor changed nothing?
Store a refusal. The Spanish AEPD's guide says that saving without selecting any cookie "equivaldrá al rechazo de todas las cookies", which means it counts as rejecting them all. The Garante says keeping the default settings is a refusal that has to be recorded.
In the Kruidvat case, a flow that recorded consent when nobody changed anything was part of the finding.
Are cookies opt-in or opt-out in the EU?
Opt-in, for anything that isn't strictly necessary. The visitor has to say yes first, by a clear action, and a default doesn't count. Strictly necessary cookies, such as a basket or a login, don't need consent.
Peeky checks the opt-in side of your settings: it reads every switch before touching it and then saves the layer untouched. How a scan works has the full path.
Filed with
The rule
ePrivacy Directive 2002/58/EC, Art. 5(3), as amended by Directive 2009/136/EC
Read the rule (ePrivacy Directive 2002/58/EC, Art. 5(3), as amended by Directive 2009/136/EC)A case
A.S. Watson Health & Beauty Continental Europe B.V. (Kruidvat.nl)
Read the decision (A.S. Watson Health & Beauty Continental Europe B.V. (Kruidvat.nl))Sources
The law
- Directive 2009/136/EC, Official Journal L 337, Art. 2(5) (replaces Art. 5(3) of Directive 2002/58/EC)
- Regulation (EU) 2016/679 (GDPR), Official Journal L 119, Art. 4(11) and recital 32
Guidance
- EDPB, Guidelines 05/2020 on consent under Regulation 2016/679, version 1.1, adopted 4 May 2020 (paras. 79 and 81)
- EDPB, Report of the work undertaken by the Cookie Banner Taskforce, adopted 17 January 2023 (paras. 9 and 10, type B: pre-ticked boxes)
- CNIL, Lignes directrices relatives aux cookies et autres traceurs (deliberation 2020-091 of 17 September 2020), paras. 50 and 51 on audience measurement
- CNIL, Recommandation sur les cookies et autres traceurs, consolidated version published 16 January 2026
- AEPD (Spain), Guía sobre el uso de las cookies, updated May 2024
- Garante per la protezione dei dati personali, Linee guida cookie e altri strumenti di tracciamento, 10 June 2021
- Datenschutzkonferenz (Germany), Orientierungshilfe für Anbieter:innen von digitalen Diensten (OH Digitale Dienste), version 1.2, November 2024
- Data Protection Commission (Ireland), Guidance note on cookies and other tracking technologies, April 2020
- Data Protection Commission (Ireland), Report on the use of cookies and other tracking technologies (sweep of 2019, published 6 April 2020)
Decisions and cases
- Autoriteit Persoonsgegevens (Netherlands), fine decision on A.S. Watson (Kruidvat.nl), 2 May 2024, in Dutch
- Autoriteit Persoonsgegevens (Netherlands), decision on objection, A.S. Watson (Kruidvat.nl), 27 May 2025: fine set at €50,000, in Dutch
- Autoriteit Persoonsgegevens (Netherlands), Boete van 600.000 euro voor tracking cookies op Kruidvat.nl (16 July 2024)
- AEPD (Spain), procedure PS/00315/2022 (EXP202206594), Freshly Cosmetics, S.L., ended by voluntary payment, in Spanish
- APD / GBA (Belgium), Litigation Chamber, decision on the merits 12/2019 of 17 December 2019, in French
- Landgericht Rostock, judgment of 15 September 2020, 3 O 762/19 (vzbv v advocado GmbH), certified copy published by vzbv, in German
- vzbv, Tracking-Cookies nur mit Einwilligung erlaubt (case note on LG Rostock, 3 O 762/19, not final)
- Garante per la protezione dei dati personali, Provvedimento del 17 ottobre 2024 [10091156] (Onlinestore S.r.l.)
- Garante per la protezione dei dati personali, Provvedimento del 17 ottobre 2024 [10091178] (Sotto il cavolo s.n.c.)
- Autoriteit Persoonsgegevens (Netherlands), Boete voor Coolblue voor het ongevraagd gebruiken van cookies (24 December 2024)
- Gegevensbeschermingsautoriteit (Belgium), decision on the merits 85/2022 on cookies on the Knack and Le Vif websites (25 May 2022, in Dutch)
Further reading
- Court of Justice of the EU, judgment of 1 October 2019, Case C-673/17 Planet49 (ECLI:EU:C:2019:801)
- Court of Justice of the EU, press release 125/19: Storing cookies requires internet users' active consent (1 October 2019)
- Court of Justice of the EU, judgment of 11 November 2020, Case C-61/19 Orange România (ECLI:EU:C:2020:901)
- IAB Europe, Transparency & Consent Framework Policies, version 2026-05-29.5.0.b (default choice 'no consent' or 'off' in the granular layer)
- IAB Tech Lab, Consent Management Platform API v2 (all consent signals false when the UI is shown)
Last checked against the source:
For information only. Not legal advice.
