Opens in a new tab
Home/Blog/Does server-side tracking bypass consent?

Does server-side tracking bypass consent?

Privacy & CompliancePhillip StemannNovember 13, 2026

Last updated: November 18, 2026

Title card: Does server-side tracking bypass consent?
In this article
Key takeaways
  • No. Moving a call to the server changes where code runs. It does not remove the need for a lawful basis for the personal data involved, or for consent where the device is touched.
  • Two laws apply: ePrivacy Article 5(3) (storing or accessing information on the device) and the GDPR (processing personal data, including sending it to an ad platform).
  • The EDPB’s Guidelines 2/2023 treat pixels, tracking links and client-side scripts that send identifiers back as “gaining access”. Most server-side setups still rely on identifiers captured in the browser.
  • A hashed email is not anonymous; per the EDPB it is still personal data if it can be attributed to a person with additional information.
  • What server-side does give you is control and one place to enforce consent, but only if you pass consent state with each event and stop forwarding when it is not granted.
  • This is general information, not legal advice. Have your DPO or a lawyer confirm your setup.

“Server-side tracking is first-party, so you don’t need a consent banner.” You hear versions of this in vendor webinars and forum threads. The regulators do not read it that way, and neither do the ad platforms’ own policies. This post separates the two legal questions people mix up, goes through the primary sources, and ends with a checklist for building consent into a server-side setup. It is part of our privacy and browsers series; the hub is Consent Mode v2 on WordPress.

This is not legal advice. I describe what published regulator guidance and platform policies say, with links, so you can read them yourself. Whether your tracking needs consent, or can rest on another basis, depends on your country, purposes and facts. For real stakes, ask a privacy lawyer or your DPO.

The short answer

Server-side tracking does not bypass consent, because consent was never a property of where the code runs. It is a property of what you do with a visitor’s device and personal data. A server-side setup can respect consent better than a pile of browser tags, or ignore it with less visibility. The architecture does not decide which.

Two questions, not one

Most confusion comes from folding two legal questions into one.

  1. Are you storing information on, or reading information from, the visitor’s device? That is the ePrivacy Directive’s Article 5(3). It is about the terminal equipment (the browser, phone or computer), and it applies to “information”, not only personal data.
  2. Are you processing personal data? That is the GDPR. It applies to any operation on personal data: collecting it, storing it, and disclosing it by transmission to another company.

The Danish authorities put the split plainly in their joint guidance (May 2025, section 1; my translation): the cookie order and the data protection rules are both technology neutral. The cookie order applies if you store or access information on the user’s equipment; data protection law applies if you process personal data in the process (Datatilsynet and the Danish Agency for Digital Government). Neither law cares whether the request leaves from a browser or a server.

Diagram of a browser, a server and an ad platform in a row. ePrivacy Article 5(3) applies at the browser step where information on the device is stored or read. The GDPR applies to every processing operation, including the server and the disclosure to the ad platform.
Where the legal obligation sits. Moving a call from browser to server changes the code path, not the duties.

ePrivacy Article 5(3): what it covers

Article 5(3) of Directive 2002/58/EC, as amended in 2009, says the storing of information, or the gaining of access to information already stored, in the terminal equipment of a subscriber or user is only allowed on condition that the user has given consent after clear and comprehensive information. It does not prevent technical storage or access for the sole purpose of carrying out the transmission of a communication, or as strictly necessary for a service the user explicitly requested (text of Article 5, in the official version on EUR-Lex).

The European Data Protection Board (EDPB) adopted its final Guidelines 2/2023 on the technical scope of this article on 7 October 2024 (PDF). They do not address when the consent exemptions apply, which depends on national law and authority guidance (paragraph 4). They answer a narrower question: is the technical operation within Article 5(3) at all? The three criteria are that it concerns “information” (not only personal data), involves a user’s terminal equipment, and constitutes storage or a gaining of access (paragraph 6).

“Gaining access” does not require a cookie

The guidelines are broad. Whenever an entity takes steps towards gaining access to information stored on the device, Article 5(3) applies; usually that means instructing the device to send the information back. Cookies, software that calls an API endpoint, and JavaScript that tells the browser to send requests all qualify (paragraphs 32 and 33). The receiver need not be the instructing party: a website instructing the browser to send information to a third-party advertising service through a tracking pixel is the EDPB’s own example (paragraph 34).

For pixels and links (section 3.1), adding tracking information to URLs or images is an instruction to send back an identifier, and for dynamic pixels the distribution of the JavaScript is the instruction. Collecting identifiers this way is a “gaining of access”, so Article 5(3) applies (paragraph 51); tracking links with an appended identifier are treated the same way (paragraph 49). The EDPB also reminds readers that Article 5(3) applying does not always mean consent is needed: each case must be assessed for consent or an exemption (paragraphs 40 and 56).

What this means for a server-side setup

Take a typical server-side conversion event. A visitor lands with a click ID in the URL, such as gclid or fbclid. A script reads that ID and the _fbp or _ga cookies and passes them to your server with the order or form submission, and your server forwards the event with those identifiers to the ad platform.

The server-to-platform hop reads nothing from the device. But the identifiers that make the event useful were collected from the device, usually by code your page delivered, and on the EDPB’s reasoning that collection is within Article 5(3). Server-side moves the last hop and leaves the first one where it was. Lifetime and consent are different questions; see our Safari ITP post for the lifetime side.

Some events may not involve the device at all: an order your WooCommerce store already holds is data your server has. Article 5(3) is a narrower question there. The GDPR still applies in full.

GDPR: lawful basis for every step

Under the GDPR, personal data includes “an online identifier”, and processing includes collection, storage and “disclosure by transmission” (Article 4(1)-(2), text on EUR-Lex). Sending an event to Meta’s Conversions API or a Google Ads upload is a disclosure by transmission. Processing is lawful only if an Article 6(1) condition applies; for marketing data the two usual candidates are (a) consent for specific purposes and (f) legitimate interests, unless overridden by the individual’s rights (Article 6). GDPR consent must be freely given, specific, informed and unambiguous (Article 4(11)), and as easy to withdraw as to give (Article 7(3)).

Which basis supports which data flow is the hard part, and where this article stops on purpose. Your own order records, a measurement tool under your control, and a hashed email handed to an ad network for its targeting are three different things, and authorities disagree on details.

What “sending hashed PII” implies

Conversion APIs invite you to send customer details such as email or phone, hashed with SHA-256. Meta’s Business Tools Terms, for example, describe hashing of contact information for matching (section 1.a).

Hashing is a security and minimisation measure. It is not anonymisation. The EDPB’s Guidelines 01/2025 on pseudonymisation (adopted 16 January 2025) tell controllers to bear in mind that pseudonymised data which could be attributed to a natural person by the use of additional information remains information relating to an identifiable person, and thus personal data, so processing it has to comply with the GDPR including Article 6 (executive summary). The same document illustrates why with a footnote: if you know a transformation is merely a SHA-256 hash of a name, you can hash all the names you already have and see which ones match.

That is exactly how ad platforms use hashed emails: they hash the emails they already hold and match. The match working is the proof that the data identifies someone. So a hashed email going to an ad platform is a disclosure of personal data, and it needs a lawful basis like any other.

Analytics exemptions: what the authorities actually say

People often counter with “analytics doesn’t need consent.” That is true in some countries under conditions, and false in others. Neither country below suggests that forwarding events to ad platforms falls inside an exemption.

France: CNIL

The CNIL’s page on audience measurement tools (4 July 2025) says trackers can be exempt from consent only if used strictly for measuring the site’s audience on behalf of the publisher and producing only anonymous statistics. They must not lead to cross-referencing with other processing or to non-anonymous data reaching third parties, and must not follow a person across sites or apps (CNIL, my summary of the French text). It recommends thirteen months as a tracker lifetime and at most twenty-five months of retention, and notes that offerings whose providers reuse data for their own account fall outside the exemption unless that reuse can be switched off.

Read that against server-side forwarding. An event sent to an ad platform for conversion measurement and ad optimisation is by design passed to a third party and used for the platform’s own purposes. That is outside the exemption as the CNIL describes it. I did not find a CNIL page that says server-side tagging changes this analysis.

Denmark: Datatilsynet and the Digitaliseringsstyrelsen

The Danish joint guidance of May 2025 says cookies and similar technologies as a rule require consent, naming visit statistics and personalised ads as examples (section 3.1, my translation). Where third parties collect information on your service you must as a rule get consent even if you have no access to what they collect (section 2.1). And responsibility for consent sits with you even when a partner or technical supplier runs the technology (section 3.6); a company generating your statistics acts on your behalf, usually under a data processing agreement. Outsourcing the plumbing, server-side providers included, does not outsource the responsibility.

What the platforms require

Google

Google’s EU user consent policy covers end users in the EEA, the UK and Switzerland. It requires legally valid consent for cookies or other local storage where legally required and for the collection, sharing and use of personal data for ad personalisation, records of consent, and clear revocation instructions (policy).

On the technical side, Google gives you fields to carry the signal. The GA4 Measurement Protocol request body has a consent object with ad_user_data and ad_personalization, each GRANTED or DENIED. If you do not specify it, Google Analytics uses the consent settings from the corresponding online interactions for that client or app instance (Measurement Protocol reference). The Google Ads API documentation for uploading click conversions describes a consent field on the conversion that specifies whether user consent was obtained for the data you upload, and says it is “highly” recommended to populate it (upload clicks guide).

If you run server-side Google Tag Manager, the Google tags in the server container are consent-aware and adjust what they send based on the consent state the web container passes along (consent mode with server-side Tag Manager). Note the limit: that describes Google’s own tags. Anything else you forward from a server is only as consent-aware as the logic you write.

Meta

Meta’s Business Tools Terms say that in jurisdictions that require informed consent for storing and accessing cookies or other information on a device, such as the European Union, you must ensure in a verifiable manner that the end user gives all necessary consents before you use Meta Business Tools to enable that storage and access (section 1.c).

The Conversions API’s Limited Data Use option (data_processing_options) is described by Meta as supporting compliance with a list of US state privacy laws; the page does not mention the GDPR or the EU (overview). Its country and state parameters are US-oriented (1 for USA, 1000 for California, or 0 for geolocation; parameters). It is not a GDPR consent mechanism.

What server-side tracking does change

None of this makes server-side a bad idea. It gives you control: you decide field by field what leaves (drop the IP address, strip query parameters), you can enforce consent in one central gate instead of in every tag, the payload is small, which helps data minimisation, and you can log what was sent where. These upsides appear only if you use them. A server that forwards everything it receives, whatever the banner said, has control and ignores the banner.

Two-column comparison. Left: what server-side can change, including control over what leaves, a single consent gate, data minimisation, first-party context and auditability. Right: what it does not change, including the need for a lawful basis, Article 5(3) at the browser step, disclosure to ad platforms, hashed email remaining personal data, and the site owner's responsibility.
What changes and what does not. Sources: EDPB Guidelines 2/2023 and 01/2025; GDPR Articles 4 and 6.

Consent signal plumbing

Treat the consent state as part of the event, not a separate system. Here is the shape of a setup that does this properly.

  1. Capture the choice and keep proof (choice, timestamp, banner version).
  2. Attach the state to every event. Treat a missing or unreadable value as “no”.
  3. Gate on the server, per destination. If marketing consent is not granted, make no call to Meta, Google Ads, TikTok, LinkedIn or any other ad destination. Dropping the call is stronger than sending it and hoping a parameter does the work.
  4. Set the platform’s own fields when you do send. For Google that is ad_user_data and ad_personalization. For other platforms I have not verified a consent field; check their current documentation.
  5. Do not use Meta LDU as a stand-in.
Flow diagram: banner records the visitor's choice, every event carries the consent state to the server, the server gates on it. Not granted: do not forward to ad platforms. Granted: forward with Google's consent fields set; Meta Limited Data Use is a US state law option, not a GDPR switch.
Consent gating for server-side events. Unknown state is treated as no consent.

What the payloads look like

A GA4 Measurement Protocol request carrying consent (field names from Google’s reference):

POST https://www.google-analytics.com/mp/collect?measurement_id=G-XXXX&api_secret=...
{
  "client_id": "123.456",
  "consent": {
    "ad_user_data": "DENIED",
    "ad_personalization": "DENIED"
  },
  "events": [{ "name": "generate_lead" }]
}

If consent is omitted, Google uses the settings from the corresponding online interactions. That helps when a browser session exists; for events with no browser counterpart, set it explicitly.

A server-side gate, in pseudocode, is boring on purpose:

function forward(event) {
  const c = event.consent;            // set by your CMP, sent with the event
  if (!c || c.marketing !== "granted") {
    log(event.id, "skipped: no marketing consent");
    return;                           // nothing leaves the server
  }
  send_to_ads_platforms(event, { ad_user_data: "GRANTED", ad_personalization: c.personalization });
}

The log line matters: a delivery log showing that a refusing visitor produced no outbound call is your evidence.

A practical checklist

Run through the figure below before you enable forwarding to any ad platform, and again whenever you add a destination. Also write the lawful basis per purpose (analytics, Meta and Google Ads are separate purposes, and your DPO or counsel should confirm each), and update your privacy notice and processor agreements so every recipient is listed.

Eight-point checklist: map identifiers, lawful basis per purpose, keep proof, send state with events, gate the forwarding, set Google's fields, skip the LDU shortcut, test by refusing.
The short version. A working checklist, not legal advice.

Where Traculo fits

Traculo forwards events from its own servers to GA4 (Measurement Protocol), Meta Conversions API, Google Ads, LinkedIn, TikTok, Snapchat, Pinterest and Plausible, with a per-event delivery log that helps with the “test by refusing” step. Its built-in data processing agreement is a contract document, not a statement that any setup is compliant. Your lawful basis, banner and consent records remain yours to design and verify. The plugin ships no consent banner and does not hold back collection itself: its server sets one _ga-compatible first-party cookie (HttpOnly, two years) unless consent is denied through the WP Consent API (statistics category) or the traculo_cookie_consent filter, and the visitor’s IP address goes to the Traculo API. The Traculo API reads no consent signal: it forwards every event to each connected destination, minus event names you ignore, so run the refusal test above.

Frequently asked questions

Does server-side tracking bypass the cookie banner?

No. A server-side call can avoid the visitor’s browser for the last hop, but the identifiers it sends were generally collected from the device, which is within Article 5(3) according to the EDPB, and the GDPR applies to the data itself. A banner is your mechanism for obtaining consent where it is needed.

Is first-party tracking exempt from consent?

Not as such. Article 5(3) does not distinguish first-party from third-party. Exemptions cover storage or access strictly necessary for a requested service or for transmitting a communication, plus narrow national analytics exemptions such as the CNIL’s, which exclude passing non-anonymous data to third parties.

Does hashing an email address remove the GDPR obligations?

No. The EDPB’s pseudonymisation guidelines say pseudonymised data that can be attributed to a person with additional information remains personal data. A hashed email that an ad platform can match is still personal data.

Can I use Meta’s Limited Data Use for EU visitors?

Meta documents Limited Data Use as supporting compliance with US state privacy laws and the page does not mention the GDPR. It is not a way to get GDPR consent.

What if a visitor declined marketing cookies but I want server-side analytics?

That depends on your country and tool. Danish guidance treats visit statistics as needing consent; the CNIL exempts strictly limited, anonymous audience measurement under conditions. Either way, do not send that visitor’s events to ad platforms.

Related reading

If you want to see exactly which events leave your site and what each ad platform answered, take a look at Traculo’s server-side event tracking. The free plan covers 10,000 events a month, enough to run the refuse-and-check test from the checklist on a real site.

ShareLinkedInXEmail

About the author

Track what matters, without the guesswork

Traculo gives WordPress sites analytics, funnels and server-side event forwarding in one dashboard. Start on the free plan with 10,000 events a month.

Try Traculo free

Related articles

Title card: Do you need a cookie banner for WordPress analytics?
Privacy & ComplianceNovember 20, 2026

Do you need a cookie banner for WordPress analytics?

Whether WordPress analytics needs a cookie banner depends on what you store and your country's rules. CNIL, Denmark, Ireland, Germany compared.

Read more
Title card: GDPR and WordPress analytics: the Danish picture
Privacy & ComplianceNovember 18, 2026

GDPR and WordPress analytics: the Danish picture

What Datatilsynet, DIGST, the CNIL and the EU-US Data Privacy Framework say about WordPress analytics in Denmark, as of October 2026. Not legal advice.

Read more
Title card: How much data do ad blockers cost a WordPress site?
Privacy & ComplianceNovember 16, 2026

How much data do ad blockers cost a WordPress site?

What ad blockers actually block, what the public numbers show, and how to measure your own analytics loss on WordPress using logs, GA4 and Search Console.

Read more