Energy-availability prediction algorithms based on neural networks and random forests have enabled energy-neutral operation on resource-constrained sensors powered by energy harvesting. Coupled with careful energy management, the energy harvesting-aware sensors can sustain perpetual uptime, provided that the energy consumption is smaller than the harvested energy. Despite recent algorithmic advancements, energy-neutral factors concerning battery characteristics and the timing of recharging and discharging processes remain to be further explored. In this article, we leverage real-world solar-energy harvesting traces to investigate the impact of these energy-neutral factors on sensor lifetime. Based on our findings, we introduce, ENORA, a novel transmission-management framework that empowers energy-neutral operation in energy-harvesting LoRa networks via embedded intelligence. It prolongs the lifetime of LoRa networks by dynamically allocating and carefully consuming energy harvested during different times of the day. The results from trace-driven simulations reveal that ENORA offers up to 2.2 times longer network lifetime than the state-of-the-art.