Transparency Features in Robo-Advice: Evidence, Mechanisms, and EU Policy Implications for Trust, Reliance, and Financial Inclusion

Digital investment advice increasingly combines automated recommendations with varying degrees of human support. Policy and design debates often assume that “more transparency”—explanations, disclosures, uncertainty communication, performance information, or human oversight cues—will unambiguously raise trust and improve consumer outcomes. This position/review paper maps the mechanisms through which transparency features can help or harm retail users, and synthesizes experimental and quasi-experimental evidence with a focus on outcomes relevant to trust/legitimacy, behavioral reliance, and access-to-human-support implications. Methodologically, we implement a Scopus-based scoping-and-extraction protocol (1,000 candidate records retrieved; 101 machine-extracted records after user-defined screening), and then restrict substantive claims to a full-text verified subset of studies for which PDFs were obtained (N=7). The verified evidence suggests systematic trade-offs: (i) providing performance information can increase perceived trustworthiness but may backfire when presented in cognitively demanding formats that reduce advice-taking; (ii) human-in-the-loop designs can increase uptake, yet may reduce decision accuracy and shift responsibility perceptions; and (iii) trust is highly sensitive to experienced accuracy, with “bad advice” producing discrete drops in trusting intentions relative to near-perfect advice, consistent with asymmetric trust updates. We integrate these findings with EU sectoral and cross-cutting governance constraints to derive testable propositions and research design implications for future survey and experimental research. The paper’s core design recommendation is calibrated transparency: layered, low-burden disclosure that improves understanding and accountability while preserving user agency and access to meaningful human support.