Build an Apartment Utility Baseline Before Capital Work

List the meters and accounts included in the review, the service each measures and the areas or equipment it serves. Distinguish a common-area electricity account from resident accounts and a whole-property water account from a partial service boundary. Record an uncertain meter relationship as a task for clarification before combining the figures.

For a project affecting only one system, keep the broader property baseline but identify the relevant measurement boundary. If the equipment shares a meter with other uses, the bill alone may not isolate its consumption. An owner can still track the bill accurately while acknowledging that more specific measurement would be needed to attribute a change to that project.

Retain the original bills and enter the service start and end dates, quantity, unit of measure and cost in separate fields. Mark estimated readings, adjustments, missing periods and account changes. Do not fill an absent period with a convenient guess and later present the series as measured data.

A lower invoice can reflect lower use, a different rate or an adjustment. A higher invoice can occur even when consumption falls. For a simple illustration, 10,000 kWh at an assumed $0.20 per kWh is $2,000, while 9,000 kWh at $0.25 is $2,250. This hypothetical example excludes fixed charges and other bill components; it demonstrates why dollars and consumption answer different questions.

A full annual cycle is a useful starting point for observing seasonal variation when complete data is available. Keep the underlying monthly or billing-period detail rather than only an annual total. Record material changes in occupancy, equipment, operating hours, leaks, construction and service boundaries alongside the figures.

If the available period was unusual, explain why and preserve it instead of quietly substituting another. A baseline chosen after the project because it makes the result look better weakens the comparison. Document the selection before work begins, including what would make the chosen period unsuitable for a later analysis.

Consider two fictional common-area electricity bills for the same meter. Period A covers 28 days and reports 2,800 kWh. Period B covers 35 days and reports 3,150 kWh. The second bill contains 12.5% more energy, but it also covers seven more days. Divide each quantity by its own service-day count: 100 kWh per day for A and 90 kWh per day for B. Average daily use is 10% lower, not 12.5% higher.

Neither result establishes project savings. Weather, occupied units, operating hours, an estimated read or another load may have changed. Daily averages correct only for period length. Keep the billed totals and the daily calculation side by side, preserve the original dates, and use the utility's service-day count or a consistently documented date convention. Do not count both endpoints one month and use elapsed days the next.

Before calculating, confirm that both periods use the same meter boundary and unit. Flag an estimated reading, overlapping service dates, missing period, reversed charge or replacement meter. Preserve the source value rather than overwriting it with a correction you cannot support. If the utility supplies a corrected statement, link it to the original and identify which version is used in the comparison.

Use the printable comparison worksheet below to document the decision. A practical outcome may be comparable for a simple daily-use illustration, usable only with stated limitations, or not currently comparable. Give the reason and next action. Do not force a percentage when the earlier quantity is zero or the units and service boundary do not match. A percentage is an output of the selected inputs, not evidence that those inputs are suitable.

For the fictional example, a defensible summary is: billed use increased from 2,800 to 3,150 kWh across periods of different lengths; average daily use decreased from 100 to 90 kWh; the cause has not been isolated. This gives a reviewer enough information to understand both numbers without presenting a favorable one as the entire story.

If a capital proposal depends on savings, preserve this preliminary comparison as a screening note and obtain the appropriate measurement-and-verification review. Do not multiply a short observed decrease by 365 and describe the result as proven annual savings. Add later evidence to the file, including findings that weaken the initial hypothesis.

ENERGY STAR's Portfolio Manager supports comparisons with a yearly baseline and other buildings. Its weather-normalized metrics help account for weather variation; the programme also distinguishes these from the ENERGY STAR score, which compares eligible buildings with peers. The official references linked below were reviewed September 24, 2026.

Normalization does not make an incomplete meter inventory complete or establish that one project caused a reduction. Retain actual use, the normalized result when available, the calculation source and known operating changes. A comparison between two properties also needs compatible boundaries and property information before a difference can be interpreted sensibly.

Record the work date, commissioning or completion evidence, measurement period and person responsible for review. Compare equivalent boundaries and relevant seasonal periods, with corrections or exclusions explained. An initial observation can identify a direction of change without being presented as a durable annual result.

A practical review can conclude that measured use fell while the share attributable to the project remains uncertain. That is more useful than an unsupported savings claim. Keep following the series, investigate unexpected changes and obtain a suitable measurement-and-verification approach when a contractual or capital decision depends on a defensible project-specific result.

• Print the utility billing-period comparison worksheet
• ENERGY STAR: benchmarking with Portfolio Manager
• ENERGY STAR: climate and weather technical reference
• ENERGY STAR: analyzing benchmarking results