Nudges rest on a dual-pillared claim: they can predictably alter human behaviour while preserving autonomy. This article challenges the compatibility of these claims. First, it identifies and synthesises the determinants of nudge efficacy into a unified framework, evaluating how they interact with the constituent elements that enable the exercise of decision autonomy. Second, it demonstrates an intrinsic, structural trade-off: determinants that maximise nudge efficacy minimise autonomy, and vice versa. This analysis offers a heretofore-elusive and novel explanation for the variable performance of nudges. Finally, it exposes the “misattribution problem” – a conceptual, methodological and policy concern where observed behaviour is mistakenly credited to a nudge rather than to latent social influence or independent personal judgment. Simultaneously engineering predetermined behavioural outcomes and preserving autonomy is theoretically intractable, suggesting the promises of nudge are fundamentally overstated.
