La théorie du nudge, en pratique

How the design of choices shapes behavior — and how to exploit that for yourself

How do nudges change behavior without restricting choices?

Nudge theory, from Richard Thaler and Cass Sunstein, holds that a nudge is any change in how choices are presented that predictably shifts behaviour without eliminating options or changing incentives — the classic examples being automatic enrolment into retirement savings, opt-out organ donation, and putting the healthier food first in the line. The honest state of the evidence is more modest than the popular account: defaults and precommitment devices replicate well and can move behaviour substantially, while the average nudge effect reported across the wider literature has shrunk considerably under meta-analyses that correct for publication bias, and several individual findings from the cafeteria and point-of-decision literature have failed to replicate. The useful reading is not that nudges do not work but that which nudge matters far more than whether you nudge, and that effects often decay without maintenance. For personal use, the strongest version is structural: change your own defaults and add friction to the behaviour you want less of, rather than relying on a reminder to out-argue you in the moment.

Nudge theory’s core claim, from Richard Thaler and Cass Sunstein’s "Nudge", is that people never choose in a vacuum — they choose inside an architecture, and that architecture biases the outcome whether or not anyone designed it on purpose. The shelf a product sits on, the box already ticked, the number of clicks to cancel: none of these remove a single option, and all of them change what most people end up doing. That is the practical payoff for behaviour change — you can design your own environment so the behaviour you want is the path of least resistance, instead of relying on motivation you may not have on the day it matters. It is also worth stating the counterweight plainly, because the field has corrected itself in public: a decade of enthusiasm produced a lot of small, fragile findings, and meta-analyses accounting for publication bias have cut the average estimated nudge effect substantially, with some celebrated cafeteria studies failing to replicate outright. Defaults and commitment devices have held up considerably better than reminders and signage. The practices below are ordered with that in mind, each with its mechanism and an honest note on how well the underlying evidence has actually held.