Validation: Quantitative
Automatic data validation and Quality Control Process (QCP)
Automatic Data Validation
The GRESB Portal has a built-in automatic validation system that flags inconsistencies as participants complete their assessment response.
Scope: What is automatically validated?
Automatic validation consists of errors and warnings displayed in the GRESB Assessment for entity-level responses, and the Asset Portal for asset-level data to ensure that all quantitative data input is complete and accurate.
The Performance Component metrics for which tests are run are the following:
Energy
GHG (Scopes 1 & 2), including specific tests on the validity of Market and Location-based GHG data
Water
Waste
Criteria: What checks are conducted?
General Requirements
Information completeness: The portal ensures mandatory evidence uploads are present, that all required open text boxes are completed, and that answers are provided for all indicators.
Data types: Certain fields must adhere to specific data types, such as numbers, percentages, text, etc.
Information accuracy: percentage values must range between 0 and 100, and metrics are restricted to absolute values (i.e., only non-negative numbers).
Asset Portal Requirements
The Asset Portal is designed to help participants report on their assets correctly. It includes a comprehensive set of validation rules for asset-level reporting, which consists of logical checks on the relationships between different data fields in the Asset Portal.
Types of asset-level validation checks:
Existence
All fields marked "mandatory" in the asset spreadsheet must be present.
Example: the fields Asset Name, Property Type, Country, State/Province, City, % of Ownership, and Asset Size (GFA) in the Asset Characteristics tab, as well as Ownership period, Status, Project Started, and Project Completed in the Reporting Characteristics tab, are mandatory.
Dates against the reporting year
Dates must fall within the reporting year (either fiscal or calendar, as specified in indicator EC4).
Example: if an asset was owned throughout the entire reporting year by an entity reporting on a Calendar Year basis, the data availability dates must fall within the range of January 1st to December 31st of the reporting year.
Location
City, state/province, and country fields must align with GRESB's third-party geocoding specification. See additional guidance on resolving asset location errors below.
Operational control logic
Asset-level data input must be directly correlated with the level of operational control selected for each asset.
Example: if the option "Whole building is Tenant Controlled" is selected, then the "Whole Building" must be selected if reporting on Energy. If "Whole building is Tenant Controlled" is selected, only the data in the corresponding Whole Building fields for energy and water are relevant for reporting. Any data in the Base Building + Tenant Space fields will be ignored.
Floor area logic
Any Maximum Floor Area field must be less than or equal to the Asset Size.
Any given Floor Area Covered field must be less than or equal to the corresponding Maximum Floor Area.
Energy-specific
Consumption and floor areas can be present for either Whole Building or Base Building + Tenant Space, but not both.
The renewable energy generated must be less than or equal to the sum of all of the consumption values.
GHG-specific
The energy and GHG tabs are logically interconnected. Therefore, the reporting level chosen to report energy consumption for an asset should be followed to report on GHG emissions.
Example: If the Whole Building is Tenant-Controlled, then only Scope 3 fields are allowed to be present in the GHG tab. Alternatively, if Whole Building – Landlord Controlled is selected, then only Scopes 1 and 2 can be present in the GHG tab.
Water-specific
Consumption and floor areas can be present for either Whole Building or Base Building + Tenant Space, but not both.
Reused and recycled water purchased off-site must be less than or equal to the sum of the reported consumption values for the same asset.
Waste-specific
If the Data Coverage is larger than zero, then the Hazardous and Non-hazardous Waste consumption fields must be present.
The sum of the proportions of waste by disposal routes must equal 100%.
Outcomes: What are errors and outliers, and how are they handled?
Errors
Errors appear in red and are generated per asset as a result of logical checks amongst the different asset-level data fields in the Asset Portal.
Participants cannot aggregate their asset data to the property subtype and country level in the Assessment Portal until all validation errors are resolved. This step is necessary to submit the Performance Component.
How to Resolve Asset Location Errors
Follow the steps below to resolve asset location errors. Note that the portal automatically re-verifies the location each time a user changes it.
Review the location errors on the Asset Characteristics page. These errors appear in red around the relevant fields.
Verify the spelling and correctness of the field(s) with errors.
Verify spelling and correctness of all other location fields, regardless of error indication.
Confirm the accuracy of the country field, even if no error is flagged.
If a street address is provided, first consider using the suggested location from the autocomplete drop-down. If error(s) persist, try removing the street address.
If the street address is not provided, add it.
If the error persists, try the following:
For State/Province: use the city name.
For State/Province: use the country name.
For City: utilize a nearby city name.
If issues persist, please contact the GRESB Helpdesk and include the hyperlink(s) or Asset ID(s) for our review.
For large numbers of location errors, GRESB recommends initially reviewing errors in the Asset Portal Data Editor. If found repetitive, then update the asset locations in bulk, using the Asset Spreadsheet import feature.
Outliers
Based on statistical modeling, GRESB identifies outliers per asset for all four key performance indicators (energy, GHG emissions, water, and waste) after all errors have been solved.
Outliers appear in grey and are observations of data points that lie at an abnormal distance from other values. This does not necessarily mean that the reported abnormal values are wrong, but participants are invited to review them to avoid any potential mistakes in the asset-level data provided.
Note that outliers do not prevent the user from submitting the Real Estate Assessment and do not affect the score (see below on Thresholds and Scoring Impact).
For information on outlier calculations, refer to the Aggregation Handbook.
Types of Outliers
There are two kinds of outliers flagged by the Portal: Intensities and Like-for-Like:
EN1
Energy
Intensity (kWh/m²)
Like-for-Like consumption change (%)
GH1
GHG
Intensity (tonnes/m²)
WT1
Water
Intensity (m³/m²)
Like-for-Like consumption change (%)
WS1
Waste
Intensity (tonnes/m²)
Like-for-like outliers are only calculated for assets and data types that are eligible for like-for-like inclusion in scoring.
Intensity outliers are only calculated if the reported floor areas covered are greater than 0.
Causes that trigger outlier detection
The two most common explanations for outliers in the past related to vacancy rates and data availability periods:
Intensity values are normalized by both vacancy and data availability.
Like-for-Like values are normalized by vacancy only. Like-for-like outliers are not normalized by data availability because an asset is only eligible for Like-for-like inclusion if the data is available for 2 continuous years.
Common mistakes that can lead to outliers
Outliers may result from reporting mistakes. When outliers are flagged, it is important to verify the following data points: Consumption/Emission, Floor Area Covered, Vacancy Rate and Data Availability.
Common reporting mistakes that result in Intensity outliers:
Using the wrong unit (e.g. MWh instead of kWh) or wrong conversion factor
Reporting full Floor Area Covered when the consumption data is incomplete; for instance, missing tenant consumption data or missing fuel consumption
Reporting on occupancy rate instead of vacancy rate
Reporting a full year of Data Availability, when the consumption data is not captured for the full reporting year
Common reporting mistakes that result in LFL outliers:
Reporting consumption data for the current reporting year and no consumption data for the previous year, with the same data coverage
Reporting a full year of Data Availability for both this year and the previous year, when the consumption data is not actually captured for a full two reporting years
Reporting on occupancy instead of vacancy or reporting the same vacancy year over year when it actually changed.
Outlier Thresholds and Impact on Scoring
Outliers are validated automatically based on thresholds.
If an outlier is detected above the fixed threshold, then the data points associated with that outlier will be included in the scoring, but will not be included in the creation of the scoring benchmark groups for data coverage and like-for-like performance change
If the outlier is substantially higher than the upper threshold (more than 1000 times greater), the data points associated with that outlier will not be included in the scoring.
The threshold for detecting like-for-like outliers varies between 20% and 30%, depending on the previous year’s consumption value.
The thresholds for detecting intensity outliers vary by data type and property type. They are based on a combination of analysis of data from previous years and market research on typical energy, GHG, water, and waste use per property type.
Intensity Outlier Thresholds
lower
upper
lower
upper
lower
upper
lower
upper
Retail: High Street
40
800
1e-5
0.4
1e-4
5
1e-5
0.4
Retail: Retail Centers: Strip Mall
40
800
1e-5
0.4
1e-4
5
1e-5
0.4
Retail: Retail Centers: Shopping Center
40
800
1e-5
0.4
1e-4
5
1e-5
0.4
Retail: Retail Centers: Lifestyle Center
40
800
1e-5
0.4
1e-4
5
1e-5
0.4
Retail: Retail Centers: Warehouse
10
400
1e-5
0.3
1e-4
2
1e-5
0.2
Retail: Restaurants/Bars
10
800
1e-5
0.4
1e-4
5
1e-5
0.4
Retail: Other
10
800
1e-5
0.4
1e-4
5
1e-5
0.4
Office: Corporate: Low-Rise Office
80
600
1e-5
0.4
1e-4
4
1e-5
0.3
Office: Corporate: Mid-Rise Office
80
600
1e-5
0.4
1e-4
4
1e-5
0.3
Office: Corporate: High-Rise Office
80
600
1e-5
0.4
1e-4
4
1e-5
0.3
Office: Business Park
1
400
1e-5
0.2
1e-4
2
1e-5
0.2
Office: Other
1
800
1e-5
0.4
1e-4
4
1e-5
0.3
Industrial: Distribution Warehouse: Refrigerated Warehouse
10
400
1e-5
0.2
1e-4
2
1e-5
0.2
Industrial: Distribution Warehouse: Non-refrigerated Warehouse
10
400
1e-5
0.2
1e-4
2
1e-5
0.2
Industrial: Industrial Park
1
400
1e-5
0.2
1e-4
2
1e-5
0.2
Industrial: Manufacturing
10
400
1e-5
0.3
1e-4
3
1e-5
0.4
Industrial: Other
1
400
1e-5
0.3
1e-4
3
1e-5
0.4
Residential: Multi-Family: Low-Rise Multi-Family
40
600
1e-5
0.3
1e-4
4
1e-5
0.3
Residential: Multi-Family: Mid-Rise Multi Family
40
600
1e-5
0.3
1e-4
4
1e-5
0.3
Residential: Multi-Family: High-Rise Multi-Family
40
600
1e-5
0.3
1e-4
4
1e-5
0.3
Residential: Family Homes
40
600
1e-5
0.3
1e-4
4
1e-5
0.3
Residential: Student Housing
40
600
1e-5
0.3
1e-4
4
1e-5
0.3
Residential: Retirement Living
40
600
1e-5
0.3
1e-4
4
1e-5
0.3
Residential: Other
40
600
1e-5
0.3
1e-4
4
1e-5
0.3
Hotel
40
800
1e-5
0.4
1e-4
5
1e-5
0.3
Lodging, Leisure & Recreation: Fitness Center
20
600
1e-5
0.3
1e-4
4
1e-5
0.3
Lodging, Leisure & Recreation: Indoor Arena
20
600
1e-5
0.3
1e-4
4
1e-5
0.3
Lodging, Leisure & Recreation: Performing Arts
20
600
1e-5
0.3
1e-4
4
1e-5
0.3
Lodging, Leisure & Recreation: Swimming Center
20
600
1e-5
0.3
1e-4
4
1e-5
0.3
Lodging, Leisure & Recreation: Museum/Gallery
20
600
1e-5
0.3
1e-4
4
1e-5
0.3
Lodging, Leisure & Recreation: Other
20
600
1e-5
0.3
1e-4
4
1e-5
0.3
Education: School
1
800
1e-5
0.4
1e-4
5
1e-5
0.4
Education: University
1
800
1e-5
0.4
1e-4
5
1e-5
0.4
Education: Library
1
800
1e-5
0.4
1e-4
5
1e-5
0.4
Education: Other
1
800
1e-5
0.4
1e-4
5
1e-5
0.4
Technology/Science: Data Center
5000
8000
1e-5
3
1e-4
5
1e-5
0.2
Technology/Science: Laboratory/Life Sciences
5000
8000
1e-5
3
1e-4
5
1e-5
0.2
Technology/Science: Other
5000
8000
1e-5
3
1e-4
5
1e-5
0.2
Healthcare: Healthcare Center
80
800
1e-5
0.4
1e-4
4
1e-5
0.3
Healthcare: Medical Office
20
600
1e-5
0.4
1e-4
4
1e-5
0.3
Healthcare: Senior Homes
80
800
1e-5
0.3
1e-4
4
1e-5
0.3
Healthcare: Other
80
800
1e-5
0.4
1e-4
4
1e-5
0.3
Mixed use: Office/Retail
10
800
1e-5
0.4
1e-4
5
1e-5
0.4
Mixed use: Office/Residential
10
800
1e-5
0.4
1e-4
5
1e-5
0.4
Mixed use: Office/Industrial
1
800
1e-5
0.4
1e-4
5
1e-5
0.4
Mixed use: Other
1
800
1e-5
0.4
1e-4
5
1e-5
0.4
Other: Parking (Indoors)
1
400
1e-5
0.2
1e-4
3
1e-5
0.2
Other: Self-Storage
10
400
1e-5
0.2
1e-4
3
1e-5
0.2
Other
1
800
1e-5
0.4
1e-4
5
1e-5
0.4
Quality Control Process (QCP)
The GRESB team conducts QCP outreach to flag inconsistencies during and after participants submit their assessments.
Scope: What does the Quality Control Process capture?
GRESB's QCP is an additional data quality validation stream aiming to enforce participants' compliance with the GRESB reporting requirements in full, and that data is reported in line with the guidance provided in the official GRESB guidance.
GRESB uses statistical modeling tools to flag additional outliers (beyond those captured automatically in the Portal). We carry these checks out on the following Reporting Characteristics metrics, which play a key role in calculating metrics such as intensities and must be closely monitored.
Gross Asset Value
Revenue
Asset floor size
Ownership values
Portfolio completeness (verification that portfolios report all assets and are not cherry-picking)
Criteria: What checks does GRESB conduct?
Descriptive Statistics
Includes measures of central tendency (mean, median, mode) and dispersion (min, max, range, variance, standard deviation, standard error). Used to detect anomalies.
An asset reporting high or low energy use while others report typical usage ranges flagged based on min/max/mean/median or standard deviation thresholds.
Interquartile Range (IQR)
Flags intensity data points that fall outside 1.5× the interquartile range of peer group. IQR does not assume a normal distribution—helps identify extreme values robustly.
A Scope 1/m2 or energy consumption/GAV emissions value 3× higher than all other assets in the same country and sector is flagged as an outlier beyond the IQR bounds.
Log-Transformed Regression
For right-skewed distributions (e.g., Gross Asset Value), log transformation is used before regression to normalize data and reduce influence of outliers.
A log-linear model uses prior year GAV and floor area to predict current GAV. Large residuals (prediction errors) are flagged for investigation.
Content & Accounting Checks
Focuses on consistency in reporting logic and alignment with protocols like the GHG Protocol—e.g., correct use of market- vs. location-based Scope 2 emissions reporting.
An entity reports Scope 2 emissions using location-based values but deducts green energy certificates—a practice only valid under the market-based method.
Energy Efficiency Intensity
Year-on-year performance trends and floor area changes are monitored to detect unexpected jumps or drops in intensity.
An asset’s energy intensity improves dramatically with no corresponding floor area change—triggering a review of the data or methodology used.
Outcomes: How does GRESB handle inconsistencies?
When GRESB detects an outlier, we contact the participant and ask them to provide additional clarification and/or use the Assessment Correction to amend their reported data.
In case of non-compliance, GRESB reserves the right to reject the corresponding component submission.
Last updated
Was this helpful?