You're the programs director at an energy utility in Lima, and you want households to cut their consumption during peak hours. The team proposes an information campaign —flyers, tips, "turn off the lights"— and the commercial team suggests paying a discount to whoever lowers their consumption. But the behavioral evidence points elsewhere: the cheapest and most durable lever isn't informing or paying, but designing the decision context. When Opower put "you used more than your efficient neighbors" on the bill, consumption dropped —no flyers, no payments.
The situation
The choice architecture is the design of the context in which people decide: how options are presented, what's set as default, what feedback they get, who they're compared with. Behavioral economics shows that this design changes behavior far more —and far more cheaply— than two classic levers. Informing (giving data and tips) barely moves behavior: people already "know," but don't act. Paying an incentive works while you pay, is expensive, and can crowd out intrinsic motivation (turning a civic duty into a transaction). Opower cut consumption with social comparison on the bill; Disney redesigned the waiting experience with FastPass instead of just reporting queue times.
Here's the trap, in two directions. The information campaign feels responsible and is cheap, but the behavior change it achieves is small: knowing isn't doing. The economic incentive does move behavior, but it costs a lot every time and evaporates —or backfires— once you stop paying. The defensible move is to design the choice architecture —social comparison, timely feedback, smart defaults—: the behavior change is large and the cost of deploying it is low, because you work with how people decide, not against it.