I examine the effectiveness of a large-scale behavioral intervention aimed at reducing peak-period electricity demand. During the summer of 2020, Tehran’s sole power distributor sent about four million blackout-warning text messages to residential and commercial subscribers, asking them to reduce consumption in the afternoon peak. I link hourly electricity use at the local distribution line level to administrative records on text messages and weather, and estimate the average treatment effect using two–way fixed–effects panel regressions, with difference–in–differences and propensity score matching as robustness checks. Across all specifications, I do not find strong, robust evidence that the text–message campaign reduced per capita electricity consumption during peak hours. Compared with the large impacts often reported in smaller behavioral studies, the results suggest that similar interventions can lose strength when scaled up to millions of households, illustrating a “voltage effect” in the context of peak-load management.