A defensible M&V plan defines the savings boundary, baseline, data, adjustments, method, operational verification, responsibilities, and acceptance process before installation changes the evidence.
Define the claim before defining the calculation
State the measure, affected equipment or systems, facility or portfolio boundary, reporting period, required result, intended users, governing agreement or program, and decision the M&V record must support. Avoid promising “energy savings” without identifying whether the claim concerns electricity, fuel, water, demand, cost, emissions, avoided use, or another stated quantity.
The M&V plan should establish who develops the baseline, collects and validates data, documents installation, approves adjustments, performs calculations, reviews exceptions, accepts results, and retains records. Contractual guarantees, incentives, regulatory programs, internal capital approvals, and operational improvement projects may require different methods and levels of independence.
Capture the baseline while it can still be observed
Before work begins, preserve utility bills, interval and submeter data, building-automation trends, equipment inventories, nameplate and setpoint information, schedules, occupancy, production or use variables, weather data, control sequences, known outages, maintenance condition, and contemporaneous operating problems. Identify meter identifiers, units, intervals, time zones, gaps, estimated values, account changes, and the system of record.
DOE FEMP guidance emphasizes that baseline physical conditions and data should be defined before installation because they can be difficult or impossible to recreate afterward. A long data history does not guarantee a representative baseline: closures, construction, abnormal weather, broken controls, occupancy shifts, tenant change, or missing end uses may require a different period, model, or documented treatment.
Choose the method that fits the measure
FEMP describes four common M&V approaches. Option A isolates the retrofit and uses measured key parameters with stipulated values for others. Option B isolates the retrofit and measures all parameters needed for savings. Option C evaluates performance at the whole-building or facility meter. Option D uses calibrated simulation when direct comparison is not sufficient or no suitable baseline exists.
The choice should follow the measure, expected savings relative to normal variation, available meters, interactive effects, uncertainty, persistence needs, reporting duration, and value of the decision. Whole-building analysis can capture combined effects but may have difficulty resolving a small measure. Retrofit isolation can provide direct equipment evidence while missing effects outside the boundary. A calibrated model depends on transparent assumptions and calibration quality.
Plan routine and non-routine adjustments
Routine adjustments account for variables expected to change regularly, such as weather, occupancy, production, hours, or another defined driver. Non-routine adjustments address changes outside the original baseline relationship, such as added floor area, tenant turnover, major equipment replacement, use changes, extended schedules, or significant operational disruptions.
The plan should identify the variables, source, expected range, calculation method, thresholds, review process, documentation, and approval authority before a disputed result. Adjustments should explain changed conditions rather than become a general mechanism for moving the result toward expectation. Preserve the original data, calculations, versions, assumptions, and disposition of each exception.
Verify installation and operation separately from savings
Installation verification asks whether the agreed measure and required components were installed. Operational verification asks whether the measure operates as intended under the relevant modes and conditions. Functional testing, controls review, setpoint confirmation, trend analysis, balancing, training, and issue resolution may contribute depending on the project.
Neither step alone establishes savings, but both protect the M&V conclusion. A calculated reduction can coexist with an unresolved control issue, and equipment can pass a point-in-time functional test without producing the expected seasonal result. Commissioning and M&V should share issue, trend, and acceptance information while retaining their distinct scope and conclusions.
Maintain the evidence through acceptance and persistence
The reporting record should show source data, quality checks, exclusions, baseline and reporting conditions, calculations, adjustments, uncertainty where required, operational findings, open issues, reviewer comments, approvals, and prior-period revisions. Automated collection can improve continuity, but ownership of meters, accounts, credentials, point names, data retention, and exception response must survive staff and vendor changes.
After acceptance, persistence may depend on continued trend review, alarm response, preventive maintenance, operator training, setpoint governance, seasonal testing, recalibration, or recommissioning triggers. The final deliverable should therefore explain not only the accepted result but also which conditions and records must be maintained for that result to remain meaningful.
Primary sources
Sources and technical references.
Source review current as of September 1, 2026.