- We find large and persistent energy savings of 11% on the target behavior in the absence of volunteer selection bias and monetary incentives

- Findings base on a dataset of over 19,000 showers from uninformed hotel guests who did not self-select into a research study

- Nature Energy Publication featured as research highlight and in several media outlets

Feedback interventions are viewed as promising behavior change strategies to foster energy conservation, and digital technologies have the potential to make them more powerful and scalable. Yet, large programs providing aggregate feedback on household energy use have failed to deliver the encouraging large savings effects reported in early studies with small samples of volunteer participants. Recently, behavior-specific real-time feedback has been identified as an alternative, more promising strategy for reducing energy consumption. Yet, it remains unclear whether the large savings effects reported can be generalized to a broader population.

We conducted a natural field experiment, presenting randomly assigned, uninformed guests at six Swiss hotels with real-time feedback on their resource consumption during showering. The devices measured the energy and water consumption of every shower taken. While guests in most rooms received real-time feedback in the shower on their resource consumption, 40% did not, serving as control group.

The treatment effects are large and significant, which clearly indicates that the digital intervention induced substantial energy conservation among the uninformed sample of participants. The specific study setting with hotel guests does not only allow us to rule out volunteer selection bias, but also financial motives for behavior change. Even in this setting the digital behavioral intervention creates large conservation effects of 11.4% or 0.21 kWh per shower. The study provides empirical evidence for digital interventions as a scalable and cost-efficient policy instrument for fostering resource conservation among the broader public.