Publication date: 23 June, 2025

Author’s note: Product teams at Meta rely on research along with other external factors to design and build products. The contributors to this project and related insights include (in alphabetical order): Bryce Brown, Briana Feng, Phil Kimura-Thollander, Maeleeke Lavan, Emily Straubel, Raleigh Upshur, Breian Witts, and Yue Zhuang.

Abstract

We wanted to understand if introducing data transparency in the product details page of the Meta Horizon app store increases Meta Quest users’ confidence that the app store gives them information they need about data collected by apps (“privacy response-efficacy”) and influences their app download decisions.

Research showed that user respondents who were exposed to transparency designs reported higher privacy response-efficacy than a control group, but only if they had not already developed a concern about data collection by apps.

Tests at feature launch showed that the transparency feature increased download of apps that collected eye tracking and face tracking data among Meta Quest Pro users.

Report

Virtual Reality (VR) headsets collect and use data, like eye tracking and face tracking, to enable apps and games on their platform to function. Some people may perceive the collection and use of this data as a reason to avoid using these apps/games (Menon et al., 2024). Our prior research suggests that experiences designed to address these concerns can build “response-efficacy,” people’s confidence that the product will effectively address their perceived problems, which may unlock use of VR products (Menon et al., 2024).

We wanted to test this research insight by building a privacy product experience in VR devices and exploring its impact on users’ privacy efficacy and behavior. The Meta Horizon app store lends itself to this exploration because it is a product context where greater transparency about data collected by apps may reassure VR users that their data collection concerns are being addressed which may facilitate app download decisions.

Research

We invited Meta Quest users to take a survey in the Meta Horizon app. We used random assignment so that half the respondents saw a ‘control’ experience and the other half saw a ‘test’ experience. The control group viewed a static screenshot of the Horizon Workrooms app in the app store in their survey.

Figure 1. Screens viewed by the control group in the survey experiment

The test group viewed a static screenshot of the same Horizon Workrooms app, but this included an App Privacy section that had information about sensor data, activity data, and info about the user collected by the app.

Figure 2. Screens viewed by the test group in the survey experiment

After viewing their assigned experience, all respondents answered the same survey questions measuring privacy response-efficacy and data collection concern (concern about data collection by app developers). See details in the Appendix.

We expected that viewing the App Privacy section would increase privacy response-efficacy, but we weren’t certain if the designs would be more or less impactful when respondents also expressed concerns about perceived data collection by apps -- that was an exploratory question.

Overall, the test group did not differ from the control group on privacy response-efficacy. But, results depended on the extent to which respondents held negative attitudes about data collection by apps. Respondents in the test group reported higher scores on privacy response-efficacy if they had “low” levels of concern (‘not at all concerned’ and ‘slightly concerned’) about data collection by apps. As respondents’ data collection concern increased from low to “medium” (‘moderately concerned’) and “high” (‘very concerned’ and ‘extremely concerned’) levels, the positive impact of viewing the App Privacy section on respondents’ response-efficacy progressively reduced. This means that viewing the new App Privacy section increased respondents’ confidence that they have the information they need only when they were not at all or slightly concerned about data collected by apps. See additional details in the Appendix.

Figure 3. Impact of transparency on privacy response-efficacy (confidence that users have the information they need, 1 = not at all confident, 2 = slightly confident, 3 = moderately confident, 4 = very confident) at low (‘not at all concerned’ and ‘slightly concerned'), medium (‘moderately concerned’), and high (‘very concerned’ or ‘extremely concerned’) levels of the data collection concern.

Product Test

The product team built the App Privacy tab in the Meta Horizon app store (in-headset, mobile app, and web). As can be seen in the image below, the feature added an App Privacy section to each app/game which informed users about what kinds of data an app’s developer may request or have access to.

Figure 4. App Privacy section in the Product Details Page of the Meta Horizon app store at the time of the product test

We monitored the impact of this new experience on usage by creating a control group that did not have access to the feature for a short period. Users were either shown the Meta Horizon app store Product Details Page including the expanded privacy section (“test” group), or one with the default privacy disclosures (“control” group). The test was conducted soon after the launch of Meta Quest Pro, an advanced VR headset that supports experiences that may use eye tracking and face tracking data. For additional details on the test method see the Appendix.

Users were highly engaged with the new experience during the test period, as approximately 79% of users who had access entered the App Privacy section at least once. The feature had a neutral impact on key product metrics overall. However, transparency was highly relevant to Meta Quest Pro users’ download of apps that collected eye tracking (gross payments in the test group were 630% of the control group) and face tracking data (test group’s payments were 797% of the control). The finding implies that transparency about data types that may be perceived as more sensitive can have a positive impact on app downloads.

Open Questions

It is unclear why data transparency increased privacy efficacy only for respondents who did not have strong concerns about data collection. Future research can explore how different amounts of transparency may be more or less effective at building privacy response-efficacy for those with and without strong concerns.

The product test shows that in some contexts (e.g., when considering use of apps that may collect data perceived to be sensitive) data transparency can facilitate app download decisions. Future experiments can explore the behavioral impact of transparency in different product contexts.

Conclusion

Taken together, these findings highlight the nuanced ways in which transparency can be reassuring and can facilitate relevant decisions for certain users and in specific contexts. This can help technology companies understand how privacy and trust play a critical role in people’s relationship with, and use of, new technology like VR products.

Appendix

Research Method and Analysis

In August 2022 we invited Meta Quest users in the United States to take an optional survey in the Meta Horizon app via an in-app notification. Random assignment was used to create two groups of respondents: (1) The test group (n = 1522) saw the Horizon Workrooms app in the app store - this included an App Privacy section that had information about sensor data, activity data, and info about you collected by the app; (2) the control group (n = 1521) saw the same Horizon Workrooms app in the app store, but without the App Privacy section. Screenshots of these experiences are provided above, in the main body of the report. After viewing the screens, respondents were asked questions measuring privacy efficacy and data collection concern:

  • Privacy response-efficacy: “How confident are you that the Quest Store gives you the information you need about the data an app developer collects about you?” [Not at all confident, Slightly confident, Moderately confident, Very confident, Extremely confident]
  • Privacy self-efficacy: “How confident are you that you can control whether an app can access your data when you use a Meta Quest device?” [Not at all confident, Slightly confident, Moderately confident, Very confident, Extremely confident]
  • Data collection concern: “How concerned are you about the data an app developer collects about you when you use a Meta Quest device?” [Not at all concerned, Slightly concerned, Moderately concerned, Very concerned, Extremely concerned]

Respondents were also asked questions unrelated to the focus of this article.

We expected groups to be similar on self-efficacy and data collection concerns because the designs did not offer options to manage apps’ access to data and because concerns are known to reflect generalized beliefs about tech companies (Hepler & Rutherford, 2020). We expected that the test group would have higher scores on privacy response-efficacy, and we explored if that effect would be moderated by levels of data collection concern.

We analyzed mean differences in survey responses between groups using an independent samples t-test. Mean differences between the test and control groups were not statistically significant for privacy response-efficacy (t(1, 1901) = -.86, p = .39), privacy self-efficacy (t(1, 1834) = -.90, p = .37), or data collection perceived problem t(1, 1792) = .58, p = .56).

We then explored if the impact on privacy efficacy depended on the extent to which respondents perceived the privacy concern. In the main body of the report, we presented average response-efficacy scores for respondents in the test and control groups who reported low, medium, and high levels of the privacy concern. In addition to simply comparing the means (as we showed in Figure 3 in the report), we also used regression analysis to formally test for a two-way interaction. Specifically, we examined the interaction effect of group and privacy concern on privacy efficacy. In the regression model, group (test or control) was entered as the predictor with the data collection concern as the moderator and response-efficacy as the criterion. We repeated the analysis replacing response-efficacy with self-efficacy as the criterion. The interactive effect of group by privacy concern was statistically significant at predicting response-efficacy (p = .01) but not self-efficacy (p = .75).

Given the statistically significant two-way interaction predicting response-efficacy, we probed the interaction further by exploring the impact of group at low, medium, and high levels of the concern. We defined these levels using percentiles for the privacy concern variable, where low = 16th percentile, medium = 50th percentile, and high = 80th percentile. These corresponded to scores at and below ‘slightly concerned’ for “low” (16th percentile), at ‘moderately concerned’ for medium (50th percentile), and at and above ‘very concerned’ for high (80th percentile). When respondents had low levels of the data collection concern, exposure to the App Privacy section increased privacy response-efficacy (B = .27, p = .005) but as respondents’ levels of data collection concern increased to medium and high levels the efficacy impact of viewing the App Privacy section weakened to non statistical significance (B = .05, p = .41 and B = -.07, p = .37, at medium and high levels, respectively)

Product Test Method and Analysis

The test was run over a 4 week period between April and May of 2023. Users were either shown the Meta Horizon app store Product Details Page including the expanded privacy section (“test” group), or one with the default privacy disclosures (“control” group).

Those in the test group were highly engaged with the privacy section during the experiment:

  • 79% of users who viewed the Product Details Page entered the privacy section at least once
    • 57% of users who viewed the Product Details Page entered the privacy section more than once
  • 21% of users never clicked into the privacy section over the duration of the experiment

When comparing the test group to the control group, impact on total download volume (total monetary value of all transactions processed by the app store in that time period) was net neutral (-0.17%, +/-.79%). However, there was an important nuance based on specific subgroups:

  • Because we were interested in the impact of transparency on download of apps that collect eye tracking and face tracking data (data types that may be perceived to be sensitive) we explored differences between the test and control groups’ app downloads among Meta Quest Pro users (the Meta Quest Pro has eye tracking and face tracking capabilities).
  • We observed an increase in gross payments for Meta Quest Pro users for both eye tracking (630%) and face tracking (797%) when comparing the test group to the control group.