Measure energy before trying to automate it

Energy automation is attractive because the goal sounds obvious: use less from the grid and make better use of solar. The difficulty is proving that an action helped rather than simply moving a number around.

Know what the measurements mean

Whole-home power, solar generation, battery flow and individual circuits describe different parts of the same system. Their signs, update intervals and units must agree before they drive an automation.

Compare the dashboard with the inverter, meter or appliance during several operating states. Check import, export, charging and standby. A reversed sensor can make a polished dashboard confidently wrong.

Create a baseline

Collect enough history to see daily and weekly patterns. Separate constant background use from flexible loads such as water heating, charging or pool equipment. Note weather and occupancy changes that could distort comparisons.

A baseline turns “the bill seems high” into useful questions: which load changed, when did it run, and was the change expected?

Automate flexible loads carefully

A flexible load has a window in which it may run and a deadline by which the outcome must be complete. Solar surplus can influence the timing, but comfort, equipment limits and the deadline remain guardrails.

For example, a load may begin after sustained export rather than a momentary spike. It may continue through a small cloud rather than cycling rapidly. If the preferred window ends, the normal schedule can take over.

Make the decision visible

A useful energy dashboard explains why a load started, how much power it used and whether the expected outcome was achieved. It should also show when an automation was unavailable or overridden.

Clarity matters because energy systems combine several independent products. When something changes after a firmware update, the record helps locate the problem.

Verify the result

Compare similar periods and consider total energy, peak demand, grid import and household comfort. An automation that increases self-consumption may not reduce total cost under every tariff. Avoid promising savings until the data supports the claim.

Keep safety outside the experiment

Do not rapidly switch equipment that is not designed for it. Manufacturer limits, electrical protection and licensed installation come before optimisation. Automation coordinates approved operating modes; it does not override protective systems.

The best energy automation is modest: accurate measurements, explainable decisions and changes that can be verified.