Shared electric scooters (e-scooter) are booming across the world and widely regarded as a sustainable mobility service. An increasing number of studies have investigated the e-scooter trip patterns, safety risks, and environmental impacts, but few considered the energy efficiency of e-scooters. In this research, we collected the operational data of e-scooters from a major provider in Gothenburg to shed light on the energy efficiency performance of e-scooters in real cases. We first develop a multiple logarithmic regression model to examine the energy consumption of single trips and influencing factors. With the regression model, a Monte Carlo simulation framework is proposed to estimate the fleet energy consumption in various scenarios, taking into account both trip-related energy usage and energy loss in idle status. The results indicate that 40% of e-scooter battery energy was wasted in idle status in the current practice, mainly due to the relatively low usage rate (0.83) of e-scooters. If the average usage rate drops below 0.5, the wasted energy could reach up to 53%. In the end, we present a field example to showcase how to optimally integrate public transport with e-scooters from the perspective of energy efficiency. We hope the findings of this study could help understand and resolve the current and future challenges regarding the ever-growing e-scooter services.