In digital markets, competition increasingly happens through the design of choices. Online platforms continuously test and refine their interfaces using behavioural data, optimising how users click, consent, subscribe and stay engaged. These choice architectures can steer users toward outcomes they may not have intended: consenting to extensive data sharing, overlooking key price information, staying longer on a platform than planned, or entering subscriptions they do not really want. Such designs often work well for companies by increasing engagement, data collection and revenues. For users, however, they can undermine informed decision-making and lead to choices that do not reflect their preferences. In digital markets, the architecture of the interface can matter as much as the product itself.
Consumer protection in the digital single market is a priority for the European Commission1 and underpins the forthcoming Digital Fairness Act. This makes the design of online interfaces a central concern for consumer protection policy. For policymakers and regulators, the challenge is determining when a commercial practice or an interface design is manipulating consumer choices.
Under EU law, a commercial practice is considered unlawful when it prevents consumers from making free and informed choices, leading them to transactional decisions they would not otherwise have taken. In the digital context, these practices are commonly referred to as dark patterns2: interface designs that manipulate, pressure or improperly steer users to accept terms, pay more or disclose data they otherwise would not.3
Beyond direct economic detriment, some designs also reduce switching and comparability, distorting competition. Detecting these effects in fast-moving markets is difficult, especially with limited users’ data, leaving an evidence gap. Closing this gap requires policy-relevant behavioural metrics that show real consumer detriment.
This is where behavioural research adds value: testing realistic interfaces under real incentives to measure whether a given design causes consumers to deviate from the choices they would have made under a fair and transparent presentation of options.4,5 To generate these metrics robustly, the Behavioural Unit of Open Evidence has already conducted more than ten experiments across multiple EU countries, each involving thousands of users, supporting the European Commission and the Consumer Protection Cooperation (CPC) Network, and will continue this work to investigate emerging practices in digital markets.
In our work so far, we have tested common design choices: asymmetric cookie banners, drip pricing vs upfront totals, discount/strikethrough claims, sponsored content/influencer labelling, opaque virtual-to-real currency conversions, and comparison tables where defaults/highlights steer attention to costlier options.
We argue that standard UX indicators (clicks, dwell time, completion) are useful diagnostics but not sufficient for consumer protection. What matters is whether users’ choices reflect their preferences under a neutral design and whether they understand the implications of those choices. Evidence should therefore link behaviour to preference alignment and comprehension, and, where relevant, capture experienced detriment (e.g., frustration, confusion). This shifts the focus from “what got clicked” to “was the decision truly informed and aligned with what the user wanted?”, generating metrics that are useful for regulators. In our work with policy makers and authorities, we operationalise economic detriment with an incentive-compatible, high-fidelity experimental design that measures whether choices align with stated preferences under a neutral design.
How our behavioural experiments work
We run online and in-person experiments with representative users, including children6 and other vulnerable groups, using high-fidelity mock-ups that isolate specific design elements (e.g., cookie options, price displays). Participants are randomly assigned to interface variants and complete realistic tasks. We record behaviour and visual saliency (paths, searches, clicks, times, final choices; eye-tracking in some studies), as well as self-reported consumer understanding and perceived frustration. Because design is the only difference, any change in decisions or comprehension across the groups of participants is attributable to that interface. To assess preference–choice alignment, we elicit preferences first, compare them with observed choices, and count systematic gaps (e.g., consenting to tracking against stated preferences) as consumer detriment.
Across studies, the core question is the same: does the choice architecture lead consumers away from the decisions they would otherwise have taken? If so, by how much, and which alternative designs reduce that misalignment? By quantifying how design shapes real decisions, behavioural science provides regulators with the evidence base needed to ensure digital markets respect consumers’ intentions rather than exploit their biases.
Work with us
If you’re preparing guidance, investigating a live case, or building a market study and need evidence beyond screenshots, we can help. We design incentive-compatible, high-fidelity experiments that replicate real websites and platforms and that can quantify detriment in ways that stand up in policy discussions and enforcement, pre-registering outcomes, powering samples appropriately (including minors and vulnerable users under strict ethics), and delivering clear effect sizes you can act on.