The Hitchhiker's Guide to Facebook Web Tracking with Invisible Pixels and Click IDs

Paschalis Bekos, Panagiotis Papadopoulos, Evangelos P. Markatos, Nicolas Kourtellis

TL;DR

This paper reveals how Facebook (Meta) uses its "invisible" FB Pixel and a one-time tag called FBCLID to track users across the web, even de-anonymizing past browsing activity and potentially tracking users indefinitely. The research highlights the significant adoption of these tracking mechanisms, particularly on sensitive websites, and their implications for user privacy.

What Problem Does This Paper Address?

The paper tackles the growing challenge of online user tracking, especially in the wake of third-party cookie restrictions. While Facebook's FB Pixel has been known to track user activity on websites, its ability to link this activity to specific users, particularly before they even have a Facebook account, and to track them persistently over time, has not been fully understood. The authors investigate how Facebook leverages the FB Pixel, in conjunction with the ephemeral FBCLID tag, to overcome limitations and build comprehensive user profiles, raising significant privacy concerns.

Methods

The researchers employed a multi-pronged approach:

  • Crawling and Analysis: They crawled a large number of websites (up to 1M, with a focus on the top 10K and a sample of 6K) to identify the presence of FB Pixel and analyze network traffic and cookies.
  • Empirical Analysis of FBCLID: They conducted experiments to understand how FBCLID values are generated and assigned, and how they are transmitted to Facebook.
  • Longitudinal Study: They used the Wayback Machine to investigate the historical adoption of FB Pixel, tracing its presence back to 2013.
  • Third-Party Sharing Analysis: They examined how websites share FB-related IDs (FBCLID and _fbp) with third-party entities.
  • Cookie Expiration Analysis: They investigated the lifespan of FB Pixel cookies (_fbp) by observing how their expiration dates are updated on websites.

Datasets

The study utilized several datasets:

  • Tranco List: A ranking of the top 1M websites, used to select websites for crawling.
  • Cyren's URL Category Check Gate: Used to categorize websites.
  • Curlie's Internet Directory: Used to identify websites in sensitive GDPR categories.
  • Wayback Machine: Used for historical analysis of FB Pixel adoption.

No new benchmarks were introduced; instead, the study analyzed existing web infrastructure and tracking mechanisms.

Key Insights and Ideas

  • FBCLID as a De-anonymizer: The ephemeral FBCLID, when combined with the _fbp cookie, acts as a powerful de-anonymization tool, allowing Facebook to link past anonymous browsing activity to a specific Facebook user.
  • Persistent Tracking Beyond 90 Days: Contrary to the typical 90-day lifespan of cookies, a significant majority of websites update the expiration date of the _fbp cookie, enabling persistent tracking indefinitely.
  • Sensitive Categories are Highly Targeted: Websites in sensitive categories like health, politics, and sexual orientation show a high adoption rate of FB Pixel, indicating extensive tracking in these areas.
  • Third-Party Sharing is Widespread: Many websites share FBCLID and _fbp IDs with numerous third-party entities, extending the reach of tracking beyond Facebook itself.
  • Historical Data Accumulation: Facebook has been building user profiles through FB Pixel since at least 2013, meaning a user's browsing history could be matched to their account from a decade ago.

Key Results

  • FB Pixel Adoption: Approximately 23% of the top 10K websites and 16.5% of sampled websites between 10K-1M rank use FB Pixel.
  • FBCLID Usage: 93% of FB Pixel-enabled websites report both _fbp and _fbc (containing FBCLID) values, enabling matching.
  • Persistent Tracking: 87.3% of analyzed websites update the _fbp cookie expiration date, allowing for tracking beyond the initial 90 days.
  • Third-Party Sharing: A median website passes FBCLIDs to 4.9-6.2 third parties, with some passing them to over 30. Google is the primary recipient, with over 61% of websites reporting FBCLIDs to its services.
  • Sensitive Categories: Websites in health, politics, and sexual orientation categories are particularly eager to adopt FB Pixel.

Limitations and Future Work

  • Limited Knowledge of Usage: The authors acknowledge they do not know how, when, or why the collected data is used by Facebook or third parties after collection.
  • Countermeasures Effectiveness: While ad-blockers and privacy-focused browsers exist, the paper notes that companies are finding ways to circumvent them, and the effectiveness of current countermeasures needs further investigation.
  • Future Work: The authors suggest further research into the effectiveness of existing countermeasures and the development of new ones to preserve user privacy in the post-third-party cookie era. They also highlight the need to quantify how effective privacy-preserving efforts are and what other methods can be proposed by the security and privacy community.

Summary updated

Abstract

Over the past years, advertisement companies have used various tracking methods to persistently track users across the web. Such tracking methods usually include first and third-party cookies, cookie synchronization, as well as a variety of fingerprinting mechanisms. Facebook (FB) recently introduced a new tagging mechanism that attaches a one-time tag as a URL parameter (FBCLID) on outgoing links to other websites. Although such a tag does not seem to have enough information to persistently track users, we demonstrate that despite its ephemeral nature, when combined with FB Pixel, it can aid in persistently monitoring user browsing behavior across i) different websites, ii) different actions on each website, iii) time, i.e., both in the past as well as in the future. We refer to this online monitoring of users as FB web tracking. We find that FB Pixel tracks a wide range of user activities on websites with alarming detail, especially on websites classified as sensitive categories under GDPR. Also, we show how the FBCLID tag can be used to match, and thus de-anonymize, activities of online users performed in the distant past (even before those users had a FB account) tracked by FB Pixel. In fact, by combining this tag with cookies that have rolling expiration dates, FB can also keep track of users' browsing activities in the future as well. Our experimental results suggest that 23% of the 10k most popular websites have adopted this technology, and can contribute to this activity tracking on the web. Furthermore, our longitudinal study shows that this type of user activity tracking can go as far back as 2015. Simply said, if a user creates for the first time a FB account today, FB could, under some conditions, match their anonymously collected past web browsing activity to their newly created FB profile, from as far back as 2015 and continue tracking their activity in the future.
BibTeX
@inproceedings{bekos2023hitchhiker,
  author = {Paschalis Bekos and Panagiotis Papadopoulos and Evangelos P. Markatos and Nicolas Kourtellis},
  title = {{The Hitchhiker's Guide to Facebook Web Tracking with Invisible Pixels and Click IDs}},
  booktitle = {The Web Conference},
  year = {2023},
  url = {https://openreview.net/forum?id=eE0Vlen00b},
}

Loading related content...

Figure