Choose the decision and boundary
Decide whether the pilot should reveal idle consumption, compare operating modes or explain a demand peak. Draw the measurement boundary and list shared loads. A meter covering two lines cannot automatically attribute consumption to one product. Record which conclusions the boundary can support and which need additional measurement.
Check dimensions before analysis
Keep power and energy distinct: kW describes a rate and kWh an accumulated quantity. For interval calculations, use the actual elapsed time and consistent units. Check the meter multiplier, transformer ratios and import/export sign with the responsible electrical team. Do not infer billing quantities from an unvalidated trend display.
Add production context
Record product, machine state, shift and good count alongside meter data. Compare like with like before attributing a change to an intervention. If output is zero, energy per part is undefined, not zero. Preserve missing intervals and counter resets explicitly rather than smoothing them into a convincing continuous line.
Review one operating cycle
Include startup, steady production, changeover and idle periods in the pilot evidence. Compare the trend with an independent reading where feasible. Agree a specific next action and who verifies it. Use the pilot to improve a measured operating practice; calculate financial savings only after validating tariffs and the baseline.
Put it into practice
Use the related tool to check your assumptions, then bring the results to your project discussion.