TL;DR
This paper investigates the impact of the GDPR on third-party tracking in Android mobile apps, analyzing nearly two million apps before and after its implementation. Despite the GDPR's intent to enhance data protection, the study found surprisingly little change in the prevalence and concentration of third-party tracking, suggesting that current enforcement and app developer practices may be limiting its effectiveness.
What Problem Does This Paper Address?
The proliferation of third-party tracking in mobile apps poses a significant privacy threat, as companies collect and share user behavioral data across various contexts to build detailed profiles. The EU's General Data Protection Regulation (GDPR), enacted in 2018, aimed to give individuals more control over their personal data and address these risks. However, there has been a lack of empirical evidence on the GDPR's actual impact on third-party tracking practices within the mobile app ecosystem. This paper seeks to fill that gap by examining whether the GDPR has measurably altered the landscape of app tracking.
Methods
The researchers employed a large-scale static analysis approach to examine the Android mobile app ecosystem. Their methodology involved:
- App Discovery and Download: They queried the Google Play Store for popular search terms to identify and download approximately two million Android apps from the UK region, both before (2017) and after (2020) the GDPR's enforcement.
- Tracking Detection: They performed automated scans of app "dex files" (compiled code) to identify all URLs associated with tracker hosts. A "third-party tracker" was defined as an entity collecting user data from first-party apps/websites to link and profile users. They focused on tracker hosts present in at least 0.1% of apps.
- Company Resolution: They identified the companies behind these tracker hosts, including their parent companies and jurisdictions, using various public information sources.
- Market Concentration Analysis: They used the Herfindahl-Hirschman Index (HHI) and other metrics (Integration Share - ISH, Prominence-Weighted Integration Share - PROWISH) to assess the concentration of market power among tracking companies.
Datasets
The study utilized two large datasets of Android apps from the Google Play Store:
- 2017 Dataset: Approximately 958,270 apps collected between August and September 2017.
- 2020 Dataset: Approximately 1,000,750 apps collected between January and March 2020.
This approach allowed for a direct comparison of the tracking landscape before and after the GDPR's implementation.
Key Insights and Ideas
- GDPR's Limited Immediate Impact: The most striking finding is the lack of significant change in the overall prevalence and concentration of third-party tracking in apps, despite the GDPR's stringent requirements. This suggests that the law's intent is not yet fully translating into practice.
- Persistence of Gatekeepers: A few large companies (Alphabet/Google and Meta/Facebook) continue to dominate the tracking landscape, maintaining high prevalence and prominence across apps. This concentration of power remains largely unchanged.
- Cross-Jurisdictional Data Flows Continue: Despite the GDPR's aim to regulate data transfers, a substantial portion of tracking companies remain based in non-EU/UK "adequate" countries (like the US), indicating that alternative legal mechanisms are likely being used to facilitate these transfers.
- App Developer Practices Lag: First-party app developers continue to share data with multiple third parties, seemingly without a significant reduction in the number of trackers, even though this practice presents compliance challenges under the GDPR.
- Potential for Future Change: While the current impact is limited, the paper notes that upcoming privacy measures from major players like Apple (opt-ins) and Google (FLOC) could signal future shifts in the tracking ecosystem.
Key Results
- Tracker Prevalence: The percentage of apps containing at least one tracker host increased slightly from 88.44% in 2017 to 91.37% in 2020.
- Number of Trackers: The median number of tracker hosts per app remained relatively stable, at 9 in 2017 and 11 in 2020.
- Company Dominance: Alphabet/Google and Meta/Facebook continued to be the most prevalent tracking companies, with Alphabet/Google slightly increasing its presence and Meta/Facebook experiencing a slight decrease.
- Market Concentration: Market concentration measures (ISH-HHI and PROWISH-HHI) showed only subtle changes, indicating a persistent, though not dramatically increased, concentration of market power among a few key players.
- Country of Origin: US-based companies remained dominant, accounting for around 90% of trackers in apps in both years.
Limitations and Future Work
The authors acknowledge several limitations:
- Partial Picture of Tracking: The analysis focused on tracking within apps themselves and did not examine data handling on tracking company servers or subsequent data sharing between third parties.
- Tracker Host Definition: The definition of "tracking" used, while based on prior research, is open to debate.
- Equal Treatment of Trackers: All tracker hosts were treated equally, without accounting for their specific purposes (e.g., advertising vs. analytics) or levels of intrusiveness.
- Android Focus: The study was limited to Android apps and the Google Play Store, while tracking is also prevalent on iOS.
Future work could involve:
- Investigating data handling on server-side infrastructure.
- Developing more nuanced definitions of tracking and intrusiveness.
- Expanding analysis to include iOS apps.
- Further empirical work to establish a causal relationship between the GDPR and observed changes.
- Monitoring the impact of new privacy measures from Apple and Google.
Summary updated