Check energy evidence before an HVAC savings claim

Define the comparison boundary first

Collect authorized meter identifiers, fuel types, units, billing-period dates, consumption values, estimated-read flags and the physical areas served. Add known operating-hour, occupancy, floor-area and equipment changes with their source dates. The deliverable is a data-readiness brief for the qualified energy reviewer, not an AI-generated savings percentage for a sales proposal.

Keep money and energy in separate columns. A tariff change can alter a bill without the same change in consumption. Likewise, a meter serving the whole building includes loads beyond the HVAC work. Ask the assistant to flag missing periods, overlaps and boundary changes rather than fill them with plausible values.

Review a fictional apparent improvement

A fictional property uses 100,000 kWh in one 30-day period and 90,000 kWh in a later 30-day period. The raw reduction is 10%. The later period also has shorter operating hours, and the records do not establish comparable weather. The correct brief reports the raw difference and those limitations; it cannot label the 10% as verified HVAC savings.

Another building shows lower utility cost but unchanged kWh. That belongs in a cost discussion with tariff evidence, not an equipment-efficiency claim. Preserve the actual readings, units and periods so the reviewer can determine an appropriate comparison and whether more information is needed.

Route normalization and attribution to qualified review

ENERGY STAR explains weather-normalized metrics for comparing building energy use over time. Its technical reference describes monthly inputs and fuel-specific relationships. AI should not imitate that model by inventing weather coefficients or assuming every building has a usable normalized metric.

Have the energy reviewer choose the baseline, adjustments and any measurement-and-verification approach appropriate to the project. Keep their approved result and assumptions linked to the underlying evidence before proposal language is released. A baseline-ready packet can support a better conversation even when no savings claim is supportable. Test the workflow using missing-meter, estimated-reading and changed-operating-hours cases, recording missed flags rather than claiming energy reductions from the administrative tool.

Review checklist

• Match meter boundary, units and comparison periods.
• Distinguish energy consumption from cost.
• Flag weather and operating-condition differences.
• Release savings language only after qualified approval.

Source context and limits

ENERGY STAR supplies benchmarking and weather context. Our fictional screening is not a Portfolio Manager calculation, engineering assessment or verified savings result. Benchmarking alone does not establish causal savings from a specific HVAC measure.