Skip to content
ClearTradeREPORT

Real reporting for the people who build, power and maintain America.

Measurement and Verification: Did the Retrofit Actually Save Energy?

How to verify retrofit savings: baseline, routine and non-routine adjustments, measurement boundary, IPMVP Options A-D, and what M&V reports must show.

By ClearTrade Report Editorial Team · Published November 17, 2026 · 11 min read

Four-step framework for measurement and verification of retrofit savings: set the baseline period, draw the measurement boundary, adjust for weather and other changes, and compare the reporting period against the adjusted baseline.
Credit: ClearTrade Report Editorial Team.

The evidence that a retrofit saved energy is a measurement and verification (M&V) package, not a lower utility bill. Ask for a documented baseline period from before the work, the adjustments that put the baseline and the post-retrofit period on equal terms (weather, occupancy, production and other changes), a stated measurement boundary with the IPMVP option (A, B, C or D) that fits it, and a written M&V plan agreed before installation, followed by savings reports that stick to it. Savings are the energy the building avoided compared with what it would have used without the retrofit under the same conditions. A bill can drop because of a mild winter, and it can rise even when a retrofit works.

Evidence label: Documentation-based guide. ClearTrade Report Editorial Team checked documents from the U.S. Department of Energy Federal Energy Management Program (FEMP), the Efficiency Valuation Organization (EVO), ASHRAE, Lawrence Berkeley National Laboratory (LBNL), the DOE Building Technologies Office, Emporia Energy and P3 International on November 17, 2026. No outside professional review was performed and nothing was tested. Companies named are documentation examples, not recommendations. This guide contains no affiliate links.

Savings are avoided energy, not a lower bill

FEMP’s M&V Guidelines: Measurement and Verification for Performance-Based Contracts, Version 5.0 says “savings cannot be measured directly because the savings represent the absence of energy or water use and related expenditures.” EVO’s IPMVP page agrees: “Energy savings in a facility cannot be directly measured because savings represent the absence of consumption or demand.”

So savings are calculated. FEMP says “Savings are determined by comparing resource use before and after the installation of ECMs” (energy conservation measures), “which may be referred to as avoided energy use.” Its Equation 2-1 reads: “Savings = (Baseline Energy − Post-Installation or Performance Period Energy) ± Adjustments.” The adjustments term is what separates verified savings from a bill comparison.

The FEMP page for Version 5.0 is dated October 14, 2024, and says “This document updates the previous version,” Version 4.0, released in November 2015. If your contract names Version 4.0, follow the contract.

Start with a documented baseline period

FEMP puts it simply: “The ‘before’ case is called the baseline.” The “after” case “is referred to as the post-installation or performance period.” IPMVP documents call it the reporting period. The baseline must be documented before the retrofit, because it cannot be recreated once the old equipment is gone.

EVO’s IPMVP Generally Accepted M&V Principles (October 2018) says the baseline period should “Represent operating modes of the facility or the equipment during a normal operating cycle,” “Include only time periods for which fixed and variable energy-governing facts are known about the facility,” and “Coincide with the period immediately before commitment to undertake the retrofit.”

For whole-building methods, FEMP says “It is best to develop models using data in whole-year sets (12, 24, 36, or 48 months).” Ask for the baseline record itself: meter data, dates, schedules, occupancy, set points and any equipment outages. EVO calls such conditions static factors, which “are assumed to remain constant.”

Routine and non-routine adjustments

FEMP says “Routine adjustments are used to account for expected variations in independent variables and energy use,” such as “weather, meals served, production, or any other factor that varies periodically.” It adds that “These adjustments often use regression analysis to correlate and adjust energy use to independent variables” to normalize energy use. In plain terms, the baseline becomes a formula (energy use versus outdoor temperature or units produced) run with reporting period conditions. The result, the adjusted baseline, is what the building would have used without the retrofit.

“Non-routine adjustments (NRAs) are used to account for unexpected changes in energy usage,” FEMP says, giving the example of a facility that “added more occupants to the space during the performance period.” Static factors “should be monitored to ensure that they are not changing,” and “The magnitude of impact on the overall project and associated cost savings determines whether an adjustment is warranted.”

EVO adds a test you can apply to any report: “The adjustment term should be computed from identifiable physical facts.” Each adjustment should point to a record, such as weather station data, a production log or a dated occupancy change.

Draw the measurement boundary

EVO says that to verify savings from specific equipment, “a measurement boundary should be drawn around that equipment.” FEMP says “The primary difference in these approaches is where the measurement boundary is drawn.” “Retrofit-isolation methods consider only the affected equipment or system independent of the rest of the facility.” “Whole-facility methods consider the total building or facility energy use,” usually at the utility meter.

Retrofit isolation can show savings for one measure while the rest of the building changes, but it can miss effects outside the boundary. EVO says “Any energy effects occurring beyond the selected measurement boundary are called interactive effects.” A whole-facility boundary captures those effects, but the savings must be large enough to stand out from normal swings in total use.

The four IPMVP options in plain English

FEMP describes the IPMVP as “a conceptual framework for measuring, computing, and reporting savings.” EVO owns the protocol, and its page says users must log in to an account to see its document library, so this section relies on EVO’s public principles document and FEMP’s guidance.

Option A, key parameter measurement. EVO: “Savings are determined by field measurement of the key parameter(s).” FEMP says “Any factor not measured is estimated based on assumptions or analysis of historical or manufacturers’ data,” and that values such as operating hours “can be estimated and then stipulated.” FEMP’s lighting example: “The key parameters are the power draws of the baseline and retrofit light fixtures.”

Option B, all parameter measurement. EVO: “Savings are determined by field measurement of the energy consumption and demand” of the isolated system. FEMP says it fits when “Energy savings values per individual measure are desired” and loads or operating hours vary. Its example is “Installation of a variable-speed drive and associated controls on an electric motor.”

Option C, whole facility. EVO: “Savings are determined by measuring energy consumption and demand at the whole facility utility meter level.” FEMP limits it to projects where “Savings are predicted to be greater than about 10% to 15% of the overall consumption measured by the utility or submeter on a monthly basis” and “Changes to the facility or its use that affect energy consumption are unlikely, or easily quantified,” during the performance period.

Option D, calibrated simulation. FEMP: “Computer simulation software is used to model energy performance of a whole facility (or sub-facility),” and “Models must be calibrated with actual interval data.” Its listed uses include “When new construction projects are involved” and “When either baseline or performance period energy data, but not both, are unavailable or unreliable.” FEMP’s Table 4-2, taken from ASHRAE Guideline 14-2023, sets acceptable calibration tolerances of ±5% mean bias error and 15% Cv(RMSE) for monthly calibration, or ±10% and 30% for hourly calibration. Here the model verifies work already done. For why a design-stage model and actual bills often disagree, see Building Energy Models vs Actual Bills: Why the Numbers Differ.

OptionBoundaryWhat is measuredWhen it fits (per FEMP and EVO)Evidence to ask for
A: Key parameter measurementRetrofit isolationThe key parameter, such as fixture power draw; other values, such as operating hours, are estimated or stipulatedOne parameter drives savings, such as a lighting retrofitField measurements of the key parameter, justification for each stipulated value, periodic inspection results
B: All parameter measurementRetrofit isolationEnergy use and demand of the isolated equipment, short-term, periodic or continuousPer-measure savings wanted and loads or hours vary, such as a variable-speed driveMeter specifications, baseline and reporting period data for the equipment, calculation method
C: Whole facilityWhole facilityUtility or main meter dataSavings greater than about 10% to 15% of measured use; facility changes unlikely or easy to quantifyWhole-year baseline, regression model and its variables, weather data source, list of non-routine adjustments
D: Calibrated simulationWhole facility or sub-facilitySimulated energy use, calibrated against metered dataMany interacting measures, new construction, or missing baseline or reporting dataSoftware name, input and output files, calibration data and statistics against stated tolerances

What the M&V plan and savings report should contain

Agree the plan before the retrofit, and check that later reports follow it. EVO’s principles list 14 required items: facility and project overview; ECM intent; selected IPMVP option and measurement boundary; baseline period, usage and conditions; reporting period; basis for adjustment; calculation methodology and analysis procedure; energy prices; meter specifications; monitoring responsibilities; expected accuracy; budget; report format; and quality assurance. Option A plans add justification of estimates and periodic inspections. Option D plans add software identification, input and output data, measured data, calibration and future changes. FEMP’s list for federal contracts also includes “Witnessing requirements and customer approval and sign-off requirements” and “Any anticipated adjustments to baseline or reporting period energy savings.”

For reports, FEMP’s post-installation report includes a “Detailed list of installed equipment” and “Expected savings for the first year.” Its annual report includes “Verified savings for the year (energy, energy costs, O&M costs, etc.)” and a “Comparison of verified savings with the guaranteed amounts for the project.” EVO’s minimum report contents include “Reporting period start and end dates,” “Energy use data,” “Verified saving calculations and methodology,” “Details of utility costs used to calculate the reported savings” and a comparison of verified savings “to the proposed savings.”

Judge the report against EVO’s principles. Procedures “should be designed to responsibly estimate savings such that they are not overstated,” and “All M&V activities should be clearly documented and fully disclosed.” A single savings total with no data, adjustments or method does not meet that standard.

Meter data, advanced M&V and submeters

LBNL’s 2017 report The Status and Promise of Advanced M&V describes methods “often referred to as M&V 2.0 or advanced M&V,” using “automated analytics that can provide ongoing, near-real-time savings estimates” and “increased data granularity in terms of frequency, volume, or end-use detail.” DOE’s Assessment of Advanced Measurement and Verification Methods (M&V 2.0) page says current evaluation, measurement and verification processes are “costly, time-consuming, and produce inconsistent results.” An automated tool still needs the same answers: which baseline, which model, which adjustments.

For small Option A or B checks, a submeter or plug-load meter can record the measured parameter. Emporia’s Vue 3 page lists “Up to 3 x 200A mains and up to 16 x 50A branch circuits” and says it can monitor usage “every second,” but describes it as “Designed for homeowners,” so confirm it suits your panel. P3 International’s Kill A Watt P4400 page says it counts “consumption by the Kilowatt-hour,” with “0.2% Accuracy” and “Max Current: 15 A,” which limits it to plug-in loads. Neither page claims IPMVP or FEMP compliance. They are data sources, not an M&V method.

Know the limits of this guide

This guide summarizes public documents. It verifies no project’s savings, and no meter or software was tested. The full IPMVP documents sit behind an EVO account login and were not read. ASHRAE Guideline 14 itself was not read; the only ASHRAE text checked was Addendum a to Guideline 14-2023, which retitles it “Measurement and Verification of Energy, Demand, and Water Savings.” FEMP writes for federal contracts; private contracts can differ. Guidance can change after November 17, 2026, so check current versions and the documents your contract names.

Primary references checked November 17, 2026: U.S. Department of Energy FEMP, M&V Guidelines: Measurement and Verification for Performance-Based Contracts, Version 5.0, and its article page; Efficiency Valuation Organization, IPMVP protocol page and IPMVP Generally Accepted M&V Principles (October 2018); ASHRAE, Addendum a to ASHRAE Guideline 14-2023; Lawrence Berkeley National Laboratory, The Status and Promise of Advanced M&V; U.S. Department of Energy Building Technologies Office, Assessment of Advanced Measurement and Verification Methods (M&V 2.0); Emporia Energy, Emporia Vue 3 product page; P3 International, Kill A Watt P4400 product page.

Frequently asked questions

Why isn’t a lower utility bill proof that the retrofit worked?

Savings are avoided energy, which cannot be metered directly. FEMP calculates savings as baseline energy minus performance period energy, plus or minus adjustments. Weather, occupancy and production changes all move a bill, so the baseline must be adjusted to the same conditions before the difference counts as savings.

What is the difference between routine and non-routine adjustments?

FEMP says routine adjustments account for expected variations in independent variables such as weather, meals served or production, often using regression. Non-routine adjustments account for unexpected changes, such as added occupants during the performance period. EVO says adjustments should be computed from identifiable physical facts.

Which IPMVP option should my project use?

It depends on the boundary and the size of the savings. Options A and B isolate the retrofitted equipment. FEMP limits Option C, the whole-facility meter approach, to projects with predicted savings greater than about 10% to 15% of measured use. Option D, calibrated simulation, fits complex projects, new construction, or cases where baseline or reporting data are missing.

What should I ask my contractor for after the work is done?

Ask for the approved M&V plan, the baseline data, and each savings report. FEMP’s annual report includes verified savings for the year and a comparison with the guaranteed amounts. EVO’s minimum report contents include energy use data, the verified saving calculations and methodology, and the utility costs used.

Disclaimer: This guide is general information based on manufacturer, standards and government documents, not professional engineering, energy auditing or contract advice for a specific building, project or product. Follow your own contract and M&V plan, and consult qualified M&V practitioners for project decisions. Guidance and standards may change. AI tools assisted with drafting. ClearTrade Report Editorial Team checked the cited sources and edited this guide.

Need a qualified professional?

Trusted Pros uses visible license-verification states and never ranks paid listings above unpaid listings.

Explore Trusted Pros