User Tracking in the Post-cookie Era: How Websites Bypass GDPR Consent to Track Users

Emmanouil Papadogiannakis, Panagiotis Papadopoulos, Nicolas Kourtellis, Evangelos P. Markatos

TL;DR

This paper reveals that many websites continue to track users through sophisticated, cookie-less methods like browser fingerprinting and ID synchronization, even when users explicitly reject cookies. The research highlights a significant disconnect between user expectations of privacy under GDPR and the actual tracking practices employed by websites, particularly less popular ones.

What Problem Does This Paper Address?

The General Data Protection Regulation (GDPR) aimed to give users more control over their data, often through cookie consent banners. However, as third-party cookies become less reliable, trackers are evolving to use more persistent and stealthy methods. This paper investigates whether websites respect user consent choices, especially when users opt out of cookies, by examining how they employ "post-cookie" tracking techniques like first-party ID leaking, ID synchronization, and browser fingerprinting. The core problem is the potential for websites to bypass GDPR consent mechanisms by using these alternative tracking methods, undermining user privacy.

Methods

The researchers developed a sophisticated crawling methodology using Puppeteer to automate browser interactions. They instrumented a Chrome browser with the Consent-O-matic extension to detect and automatically handle cookie consent forms, allowing them to simulate three user actions: "Accept All," "Reject All," and "No Action." They then collected extensive data, including HTTP requests/responses, cookies, and JavaScript execution profiles. For detecting browser fingerprinting, they focused on analyzing JavaScript function calls indicative of fingerprinting libraries, rather than just native browser methods, to reduce false positives. For ID synchronization and leaking, they performed offline analysis of network traffic, examining URLs, request bodies, and headers for unique identifiers being passed between domains.

Datasets

The study crawled approximately 850,000 websites from the Tranco list. From this, they identified 27,953 websites with cookie consent management platforms (CMPs). The collected data for these sites, across the three consent actions, amounted to 108 GB. The paper does not introduce a new benchmark dataset but rather uses a large-scale crawl of real-world websites to analyze existing tracking practices.

Key Insights and Ideas

  • Consent is Largely Ignored: The most striking finding is that a significant portion of tracking activities, particularly first-party ID leaking, occurs before users even have a chance to make a choice or after they explicitly reject all cookies.
  • "Reject All" Can Intensify Tracking: Counterintuitively, rejecting all cookies sometimes leads to more third-party tracking activities, especially ID synchronization, suggesting that opting out might trigger fallback tracking mechanisms.
  • Browser Fingerprinting is Persistent: Websites largely disregard user consent for browser fingerprinting, performing it regardless of the user's choice. This technique is particularly concerning as it's stateless and harder for users to control.
  • Popularity Matters (for Disregard): Less popular websites are more aggressive in disregarding user consent choices and engaging in first-party ID leaking and third-party ID synchronization. This suggests a correlation between a website's market position and its adherence to privacy principles.
  • Geographic Discrepancies: Websites in certain European countries (e.g., France, Denmark) show better adherence to user choices regarding first-party ID leaking compared to others, indicating varying levels of GDPR enforcement or implementation.

Key Results

  • First-Party ID Leaking: Over 75% of first-party ID leaking activities happened either before users made a choice or when they rejected all cookies. Specifically, 14,238 websites leaked IDs before consent, and 15,334 did so when users rejected all cookies.
  • Third-Party ID Synchronization: Websites with embedded third-parties engaged in an average of 3.51 synchronizations before user consent, and this number increased to 3.91 when users rejected all cookies.
  • Browser Fingerprinting: 279 websites performed fingerprinting before user consent, and 285 did so when users rejected all cookies. A significant 73.5% of fingerprinting websites did so regardless of the user's consent action.
  • Popularity Impact: Less popular websites showed a trend of being more aggressive in disregarding user choices. For instance, popular sites (top 50K) leaked first-party IDs before user choice 67% of the time, while less popular sites (400K-450K) did so 77% of the time.

Limitations and Future Work

The authors acknowledge that their fingerprinting detection might miss minified or obfuscated scripts. They also note that while their methodology can identify potential GDPR violations, a definitive legal judgment would require a case-by-case analysis, as websites might have other legal bases for data collection beyond explicit consent.

Future work could explore:

  • How third-parties should connect into CMP prompts.
  • Whether certain third-parties intentionally act only on "reject all" options and why.
  • The comparative effectiveness of different CMPs in ensuring GDPR compliance.
  • The attribution of privacy violations: is it the website, the CMP, or the third-party's fault?
  • Investigating the dynamics of tracking and how they evolve over time.

Summary updated

Abstract

During the past few years, mostly as a result of the GDPR and the CCPA, websites have started to present users with cookie consent banners. These banners are web forms where the users can state their preference and declare which cookies they would like to accept, if such option exists. Although requesting consent before storing any identifiable information is a good start towards respecting the user privacy, yet previous research has shown that websites do not always respect user choices. Furthermore, considering the ever decreasing reliance of trackers on cookies and actions browser vendors take by blocking or restricting third-party cookies, we anticipate a world where stateless tracking emerges, either because trackers or websites do not use cookies, or because users simply refuse to accept any. In this paper, we explore whether websites use more persistent and sophisticated forms of tracking in order to track users who said they do not want cookies. Such forms of tracking include first-party ID leaking, ID synchronization, and browser fingerprinting. Our results suggest that websites do use such modern forms of tracking even before users had the opportunity to register their choice with respect to cookies. To add insult to injury, when users choose to raise their voice and reject all cookies, user tracking only intensifies. As a result, users' choices play very little role with respect to tracking: we measured that more than 75% of tracking activities happened before users had the opportunity to make a selection in the cookie consent banner, or when users chose to reject all cookies.
BibTeX
@inproceedings{papadogiannakis2021user,
  author = {Emmanouil Papadogiannakis and Panagiotis Papadopoulos and Nicolas Kourtellis and Evangelos P. Markatos},
  title = {{User Tracking in the Post-cookie Era: How Websites Bypass GDPR Consent to Track Users}},
  booktitle = {The Web Conference},
  year = {2021},
  url = {https://openreview.net/forum?id=shYUt6IKph},
}
Figure