ENERGY STAR Portfolio Manager data quality: the checks that keep scores and reports defensible

Every large-building benchmarking rule in Canada that matters today runs on the same tool. Ontario's Energy and Water Reporting and Benchmarking program, Toronto's by-law, Montréal's GHG disclosure by-law and most US city ordinances all take their submission from ENERGY STAR Portfolio Manager. That makes Portfolio Manager data quality the shared failure point: a gap in one meter, a bill entered twice, or a default value nobody replaced produces the same wrong number in every report that reads from the property.

The tool knows this, which is why it ships with a Data Quality Checker and a long list of alerts. Most teams run the checker the week before the deadline, fix what is red, and submit. The alerts are useful, but they are symptoms. The causes sit upstream, in how the utility data was collected and keyed, and that is where the fix has to happen.

This guide covers what the checker tests, which alerts trace back to which defects in the source bills, how errors move the score and the EUI, and a checklist to run before the deadlines. For property setup, start with Portfolio Manager in Canada.

Why data quality is the shared failure point

Ontario requires owners of buildings 50,000 square feet and larger to report energy and water use by July 1 each year under O. Reg. 506/18, entering data in Portfolio Manager and submitting through the EWRB page; properties of 100,000 square feet or more need a professional verifier in the first reporting year and every five years after. Toronto's Municipal Code Chapter 367 by-law uses the same tool, with 2025 data due to the City on July 2, 2026 and buildings of 10,000 to 49,999 square feet joining from July 2, 2027. Montréal's GHG disclosure by-law takes monthly whole-building energy data through the same tool. Full detail on the Ontario mechanics is in the EWRB reporting guide.

Three programs, one property record. If the record is wrong, it is wrong in three places at once. NRCan's Canadian adaptation of the tool, launched in 2013, uses more than 150 Canadian weather stations and gives six building types a Canadian 1-100 score, which means the weather normalization is only as good as the postal code and the bill dates you entered.

What the Data Quality Checker actually does

ENERGY STAR describes the checker plainly: it runs a set of basic data checks on a property to help identify possible data entry errors and to see whether the building differs from typical operational patterns. There are more than 70 data quality checks performed in Portfolio Manager, and ENERGY STAR publishes the complete list of alerts as a downloadable file regenerated from the tool every 24 to 36 hours. Its Portfolio Manager 201 training walks through running the checker alongside editing property data and correcting property use details.

Two glossary definitions do most of the work. A gap is a date not covered by any of your bills. An overlap is the same date covered by more than one bill. Both come from the bill dates you entered, not the usage figures, which is why plausible-looking totals can still fail. The third concept is the 12 full calendar months rule: a full month includes the first and last day of that month, so if your bills run mid-month to mid-month you need 13 bills to make up 12 full calendar months. The data collection worksheet lists 12 consecutive months of energy data among the requirements for every property, alongside gross floor area, year built, occupancy and number of buildings.

The alert families worth knowing

The published list is long, but the alerts cluster into a handful of families. The glossary calls the summary versions alert metrics and says they exist to spot check properties for the most common errors that cause metrics to show as N/A. In practice you will meet these groups.

• Meter coverage: gaps, overlaps, and meters with fewer than 12 full calendar months in the reporting period
• Estimation and defaults: bills flagged as estimated, and property use details carrying the Default Data Flag because hours, workers or computers were never replaced with real values
• Property use details: areas that do not add up to the gross floor area, or values outside the range the tool expects for that use type
• Score eligibility: anything that stops the 1-100 score from calculating, which is usually one of the above

Where the alerts really come from: the utility data

Almost every alert is a downstream symptom of one of five defects in the bills and meter files. The Estimation flag is a good example: the tool records that an entry was estimated, but whether it is ever replaced by an actual read depends on your upstream process. Clear only the alert and you will clear it again next year.

The unit row is the quiet one: a gas bill can show volume, energy, or both, and a portal export can switch between them without warning. The general treatment of these failure modes is in utility data quality for energy programs. The checker cannot see any of them directly; it only sees their consequences.

How wrong data moves the score and the EUI

EUI is a simple ratio: ENERGY STAR defines it as total energy consumed in one year, in kBtu or GJ, divided by gross floor area, and notes that both site and source EUI are available while EPA relies on source EUI as the basis for the score. That arithmetic tells you the direction of each error. A missing meter or a gap lowers the numerator, so the property looks better than it is. An overlap or a duplicated bill raises it. A floor area that is too small inflates EUI; one that includes space the meters do not serve deflates it.

The score adds a second layer. EPA states that all of the calculations are based on source energy and account for the impact of weather variations as well as changes in key property use details, that a property is compared with buildings nationwide that share its primary use, and that 50 is median performance while 75 or higher is a top performer. So property use details are inputs to the expected energy, not descriptive fields: leave a default worker count or operating hours in place and the tool compares your actual energy against the wrong expectation. The guide to improving the score covers the levers; the point here is that a data error can move the score in either direction, and neither is defensible when a verifier asks for the bills.

The pre-submission checklist

Run this in the order shown, at least a month before the June 30 and July 1 dates, because the fixes that involve a utility take weeks.

1. Confirm the meter inventory: every account, meter and commodity serving the property, including tenant meters where whole-building data is required
2. Check each meter for gaps and overlaps and confirm 12 full calendar months (13 bills for mid-month cycles)
3. Replace every estimated entry with the actual read where the true-up bill exists
4. Verify units per commodity against the bill itself, not the portal export label, and clear every Default Data Flag with real values
5. Reconcile gross floor area and property use areas to the current drawings or lease schedule
6. Run the Data Quality Checker, then compare the score and EUI with last year and explain any move larger than you would expect
7. Archive the bills behind every entry so the verification package is ready without a second collection exercise

How automated data exchange changes the cadence

Portfolio Manager's web services API lets a utility or data provider push consumption data directly into a property record, usually as aggregate whole-building consumption rather than meter by meter. Where that exists, the annual scramble becomes a monthly check. Where it does not, which is still most Canadian properties and most gas, water and steam accounts, the same cadence can be built from bills and portal exports; the mechanics are in automating Portfolio Manager updates.

This is where MartinAI fits. It reads every field on every bill and meter file for every commodity, checks period continuity, units, estimation status and meter coverage as the data lands, and delivers a validated monthly dataset with each figure tied to its source document. The checker then confirms what the upstream validation already established.

Data quality in Portfolio Manager is not a feature of the tool. It is a property of the utility data you feed it, checked before entry and traceable after. Get that right and the alerts, the score movements and the verifier's questions become predictable.