In February 2020, we published our review of online targeting, which included a number of recommendations to government, regulators and industry, regarding how data is used to shape the online experience. In the final report, we recommended that online platforms should improve the information and controls they offer to users. To further strengthen the evidence base, we partnered with the Behavioural Insights Team (BIT) to find ways to empower people to make ‘active’ online choices. We defined active choices as choices that reflect users’ wishes without obstruction, which are based on an understanding of the likely consequences. The aim of the project was to:
- Identify ways to design online choice environments that empower people to set user controls in ways that align with their preferences.
- Provide firms operating online with examples of evidence-based tools and techniques which would help them to design interfaces that empower users to make active choices.
We shared a progress update in November 2020, and published an interim report in June 2021, which detailed the findings of the first of three experiments. Today, BIT has published the final report which includes the results from all of the experiments, and sets out the implications of this work for policymakers, regulators and industry. The accompanying technical report provides additional information on the methodology used to run the experiments and findings from them.
Policymakers, regulators, industry and civil society are all considering how to empower users online. A number of policy initiatives, including the Draft Online Safety Bill, Information Commissioner’s Office’s (ICO) Age Appropriate Design Code and establishment of the Digital Markets Unit, have recognised the central role that empowering users through active choice can play as part of a broader approach to digital regulation. The Competition and Markets Authority (CMA) called for a “fairness by design” duty in its market study of online platforms and digital advertising. Our work with BIT also coincided with announcements by various prominent technology firms to consider user design in how people make decisions about their privacy and data. In January 2021, Apple announced the introduction of opt-in data consents for apps in iOS14, and in May 2021, Google launched features to improve the interface for safety and privacy controls as part of Android 12. Last year, Facebook published a white paper on its work with regulators, governments and civil society to improve the communication of privacy information. It has also recently introduced new tools to give users’ greater control over their News Feed. However, further action is needed to enable active choices online.
Research undertaken by BIT
In this project, BIT carried out the following research activities:
- Desk research, first-hand website audits and expert interviews with people working in the technology industry. These activities helped to identify the existing barriers to active choices in these online environments (such as high prevalence of defaults, and poor timing of prompts).
- Prioritised areas for improvement by selecting three common user contexts, including: a smartphone; a web browser and a social media platform.
- Created alternative interface prototypes for each context, drawing on behavioural science principles identified in the literature review (which can be found on p.12 of the final report), that we expected would promote active choices. Examples of these alternative interface prototypes included: a slider mode design (which allowed the user to select an option along a spectrum); a private mode design (which bundled choices together into two options); and a trusted third party design (which asked users to delegate choice to a third party organisation).
- User testing, which enabled iteration of the prototypes before robust quantitative testing.
- Ran three online experiments, each with c. 2,000 participants, to test how our alternative interfaces performed against the controls.*
When testing the behaviourally informed prototypes against the controls, BIT used three measures identified as the components of active choices in the exploratory research phase of this project:
- Task accuracy: Could participants adjust settings to match the preferences of a fictional persona?
- Understanding of consequences: Could participants correctly indicate the implications of their choices?
- Feelings of control: Did participants feel in control of their experience?
While an online experiment with simplified interfaces and a limited number of choices cannot fully replicate the experience of using technologies in everyday life, there were some significant findings that point to the benefits of active choice design:
- Some prototypes outperformed the control design across all three measures in both the smartphone and web browser experiments.
- Simplifying and bundling privacy settings better enabled users to adjust settings to match the preferences of a fictional persona, in these two experiments on average. Their understanding of consequences also improved and their feelings of control either improved or did not change.
- There was no ‘one size fits all’ solution. No single design significantly outperformed all of the others across the three main outcomes. The performance of designs also varied by persona chosen.
We would encourage policymakers, regulators and industry to consider the implications of this work and build on our findings.
Further research could be done to test some of the design ideas in real life settings. A promising avenue of further exploration is looking at the best ways to combine or bundle options to reduce the complexity for users, as well as the optimal moments in time and frequency to engage with users. It would also be valuable to investigate the differences across participants who chose different personas, and to develop designs with the most vulnerable groups of users in mind. Importantly, further research is needed on delegating decisions to trusted third parties, that engages regulators, as well as leading consumer protection and civil society organisations.
Additionally, regulatory frameworks for online interactions should consider incorporating active choice requirements where appropriate, as obtaining consent is not necessarily a marker that the user has made choices in line with their preferences and understood the consequences. Meanwhile, this work should encourage industry players to consider applying the same level of service optimisation for active choices in settings as they do to other parts of their platform design. It will be important for companies to develop, adapt and test their interfaces to ensure they better enable active choices, and for them to share updates on their development and findings.
If you would like to talk to us about this work in more detail, please get in touch with us at firstname.lastname@example.org.
*We chose a number of common existing interfaces as a baseline. The purpose of this was to have a realistic control. We then simplified and modified the control interfaces to allow for fairer comparison with our designs. Our mock-up designs were not endorsed by any organisations.