How to Build a Useful Utility-Usage Baseline

A baseline is a comparison range, not one magic average

“Normal usage” changes with weather, occupancy, operating schedules and equipment. A useful baseline represents conditions similar to the period you are evaluating.

Start with clean intervals

Choose readings that belong to the same meter, use the same unit and multiplier, and have reliable timestamps. Exclude meter replacement boundaries and identify estimated values.

Calculate units per day when interval lengths differ. Then group periods by meaningful context:

• Warm versus cold season
• Occupied versus vacant
• Weekdays versus workshop operating days
• Before versus after major equipment changes

Use a range

Suppose three comparable periods average 4.7, 5.1 and 5.4 units/day. A baseline range of roughly 4.7–5.4 communicates normal variation better than declaring exactly 5.07 to be normal.

If the next comparable interval reaches 7.2 units/day, investigate—but first verify the reading and context.

Avoid weak comparisons

• One unusually low holiday period is not a baseline.
• A partial 10-day cycle should not be compared by total with a complete 32-day cycle.
• Home and workshop meters should not share a baseline.
• A weather-sensitive meter needs seasonal context.
• A tariff change affects cost, not the physical unit baseline.

Let the baseline evolve carefully

Equipment upgrades, household size or business hours can create a new stable pattern. Do not immediately absorb an unexplained spike into “normal”. Require several validated intervals and document the known change.

MeterPilot anomaly detection works best when each meter has enough trustworthy history. When evidence is sparse, the honest result is insufficient confidence rather than a dramatic warning.

Turn comparison into action

Use the baseline to decide whether to verify a reading, shorten the next interval, review operating changes or seek professional help. It is a decision aid, not proof of a fault or a guaranteed future bill.

Prefer the middle of several periods

One extreme interval can pull an arithmetic average away from typical use. With comparable daily rates of 4.8, 5.0, 5.2, 5.3 and 11.9, the middle value is 5.2 kWh/day while the average is 6.44. Investigate the 11.9 interval instead of allowing it to redefine normal immediately.

You do not need advanced statistics. Keep a visible range, identify the middle few observations, and retain notes for known events.

Require enough coverage

A baseline based on two readings has one interval and is fragile. Three to six comparable intervals are more useful; seasonal equipment may require a full year before summer and winter can be separated honestly.

When MeterPilot says coverage is partial or an interval crosses the selected range, preserve that qualification. A clean-looking chart does not create observations that were never taken.

Reset deliberately, not silently

After a permanent occupancy or equipment change, keep the old baseline labelled by period and build a new one. Historical comparison then explains the shift rather than rewriting the past.