Industry accounts for more than one third of global final energy consumption, yet energy management still relies on aggregate, retrospective metrics: the monthly Specific Energy Consumption (SEC), ISO 50001 energy performance indicators (EnPIs) and energy-augmented OEE all report a single ratio per period. None attributes energy to the machine state that generated it, so they identify that waste exists but not where it originates, and they cannot drive a control action. This work develops and validates the Dynamic Specific Energy Consumption (SECdin), which addresses this gap on three fronts: diagnostic granularity, decomposing consumption per asset into active operation, idle standby and low-power terms, making waste auditable; real-time suitability, being updated at each acquisition and converging within the shift rather than at period close; and control integration, since its standby term parameterises the switching decision itself. Grounded in Design Science Research, a three-layer IoT architecture (edge current sensing, fog/cloud state correlation, application rule engine) was validated by Discrete-Event Simulation of a serial three-station manufacturing cell over 8-hour shifts, with 30 paired replications and paired t-tests at the 5% level. Three scenarios were compared: passive baseline, fixed 30-second timer and SECdin-guided per-machine thresholds derived from each asset's recoverable standby waste. Automated low-power switching reduced consumption by 12.38% (32.86 kWh per shift; p < 0.001), scaling monotonically from 2.22% to 47.80% with idleness. SECdin-guided control matched the fixed-timer saving with 29.2% fewer switchings and raised the energy recovered per switching from 0.126 to 0.156 kWh, mitigating thermal wear — a gain unattainable without state decomposition. SECdin converts aggregate energy accounting into auditable, asset-level, real-time control, shifting industrial energy management from retrospective reporting to preventive and verifiable action.