Before claiming energy savings, define the comparison
Evidence and practical analysis | Logic Foundry | October 12, 2026. The metric definitions below come from U.S. EPA ENERGY STAR Portfolio Manager guidance. The review method is our proposal, illustrated with a fictional example, not a report of measured savings or a completed experiment.
1. What does a lower total establish?
Imagine a fictional property whose reported electricity use falls after an operating schedule is changed. During the same period, part of the building becomes vacant. The lower total is worth investigating, but that observation alone does not separate the effect of the schedule from the effect of the vacancy.
The useful question is narrower than whether the building improved: what comparison would support the conclusion being considered? A facilities team discussing an operational adjustment, an owner reviewing a retrofit and a designer assessing a proposed control sequence may need different evidence. State the question before selecting the most attractive chart.
2. Different metrics answer different questions
The U.S. EPA's ENERGY STAR Portfolio Manager glossary, accessed October 12, 2026, distinguishes site energy from weather-normalized site energy. The latter estimates use under average weather conditions. It also defines site energy use intensity as site energy divided by property floor area. These are tool definitions, not measured results for a Logic Foundry customer or a national savings estimate. [1]
The same glossary separately describes energy adjusted to the current time period, which accounts for operating conditions for properties eligible for that metric. Our inference from these distinct definitions is that a weather-normalized value should not be described as if every operating change had been held constant. Use the documented meaning of the selected metric and check its applicability to the property. [1]
No portfolio performance statistic is being claimed here. The cited guidance is from a U.S. federal programme and describes Portfolio Manager metrics. It does not establish that a particular comparison is sufficient to attribute savings to a project, nor does it certify an unrelated software product.
3. Write the comparison in one paragraph
Our proposed review starts with a short statement identifying the property, energy stream, reporting periods, metric and intended conclusion. Add the reason those periods are being compared. A reviewer should be able to tell whether the question concerns consumption, spending or the possible effect of a specific intervention.
For the fictional property, the statement might say that the team is reviewing electricity use before and after a schedule adjustment while separately recording the vacancy. It would not call the entire difference schedule savings. That restraint leaves room to investigate competing explanations without discarding the observation.
Keep the measurement boundary visible. Note which meters and spaces are included, whether the same boundary applies in both periods and where information is incomplete. If equipment or a tenant has moved outside that boundary, record the change rather than let it disappear inside a total.
4. Keep a timeline beside the chart
As a practical proposal, build a timeline of events that could matter to the comparison: operating hours, occupancy changes, equipment replacement, meter changes, unusual closures and the intervention itself. Mark an event as confirmed, estimated or unknown, and retain the source of that judgement.
A timeline is not a mathematical adjustment model. Its purpose in this review is to prevent a confident explanation from overlooking an obvious change in circumstances. A qualified analyst may need a more formal evaluation for an investment decision or a contractual performance claim.
Review the underlying records before adding explanatory complexity. Check reporting dates, units, gaps and estimated readings. Separate a corrected data entry from a change in building operation. If the available information cannot support the intended conclusion, narrow the conclusion and describe the missing evidence.
5. Try a comparison review without changing operations
Readers could select one existing, non-sensitive energy report and ask two reviewers to write down what they believe it demonstrates. Give them the same metric definition and boundary description, then compare their answers. Do they agree on what was observed and what remains unexplained?
Next, add the operating timeline and repeat the discussion. Record whether the additional context changes the conclusion, raises an unresolved question or makes no difference. Use historical records; this exercise does not require changing equipment schedules or live control settings.
The hypothesis is that making the comparison explicit could reduce unsupported attribution in portfolio reporting. This exercise tests the clarity of a report and the reasoning of its reviewers. It cannot validate a savings model or establish annual project savings, and Logic Foundry has not conducted it.
6. Design reports around the claim they support
A proposed future report could show the measured series, the selected comparison and a concise account of known changes together. It should distinguish an observed difference, an analytical estimate and a verified project outcome rather than present them under one undifferentiated improvement label.
Two discussion prompts can guide a portfolio meeting: Which operating change would most alter your interpretation of this chart? What additional evidence would be needed before attributing the difference to the intervention? Those questions help turn a favourable number into an investigation with a defined next step.
CORE-FM currently supports properties, assets, work orders, preventive maintenance and sensor monitoring. Logic Foundry's confirmed research includes energy optimisation and future reporting and decision tools. The review approach above informs that discussion; it does not claim a shipped weather-normalization feature, verified savings model or measured energy reduction.