Data Coverage & Completeness – SIERA

Introduction

Data Quality is represented by two separate metrics: Coverage, which shows how much of the expected area your data is covering: and Completeness, which shows how much of the excepted time you've collected data for. These metrics are shown as two sperate charts
This guide is a detailed explanation of the logic and calculations underlying the coverage and completeness metrics, including worked examples for various cases.

Methodology

Area-based Coverage

Data coverage tells you how much floor area consumption data was collected for, as a proportion of the area that data “should” have been collected for. When considering multiple utilities at an asset, or multiple assets across a fund, the covered and expected areas are simply added up across the assets. Drilling down to asset level allows you to see exactly which meters or utilities are not collecting data from the full expected floor area.

The set of assets and utilities being included in this calculation differs according to where you’re seeing the visualisation:

At fund level, all the utilities expected for all assets in the fund are included by default. On the fund Data Quality page, this can be filtered to show coverage for particular utilities, for assets under certain managing agents, or for meters which are landlord- or tenant-procured.
At asset level, all the utilities expected at the asset are included by default. On the asset Data Quality page, this can be filtered to show coverage for particular utilities, and for meters which serve certain areas of the building.

Worked Examples of Coverage

To see a concrete example, let's understand this calculation for an asset where we expect 3 utilities, Electricity, Gas and Water. In SIERA terms, this means District Heating, District Cooling and Oil are all set to be ignored.

Asset Ref	Total floor area (GIA)	Net lettable area (NLA)	Common parts area (CPA)
Asset1	10,000 m²	8,000 m²	2,000 m²

Suppose we have two electricity meters each serving 5,000 m², which both collected data, and two water meters each serving 5,000 m², one of which has no data in the period. There are no registered gas meters, even though SIERA is expecting this utility.

Meter Ref	Utility	Procured by	Area Served (m2)	Data recorded in period?
Elec_Meter_1	Electricity	Landlord	5000	Yes
Elec_Meter_2	Electricity	Tenant	5000	No
Water_Meter_1	Water	Landlord	5000	Yes

This example above would see the data quality card complete this formula and show these values;

Area covered is the sum of NLA served for all meters with data collected:

Electricity	=	5,000
		+
Water	=	5,000
		+
Gas	=	0
		= 10,000 m² floor area covered by consumption data

By default the total area expected is the GIA of the asset, or the NLA for Fuels and Thermal utilities (in line with GRESB logic, and EVORA gap-filling logic).

Utility	Default Expected Area	For this example
Electricity	Asset GIA	10,000
Water	Asset GIA	10,000
Gas	Asset NLA	8,000
District Heating	Asset NLA	N/A
District Cooling	Asset NLA	N/A
Oil	Asset NLA	N/A

In this case, we have:

Electricity	=	10,000
		+
Water	=	10,000
		+
Gas	=	8,000
		= 28,000 m² meter floor area expected

Subtracting the covered area from the total, we have 18,000m² floor area unaccounted for by consumption data.

This corresponds to the water meter which hasn’t collected data, and the missing gas meters. With these numbers found, the calculation for % floor area coverage is:

10,000 ÷ 28,000 = 36% Data Coverage (SIERA will always round to the nearest whole number).

Complex Metering Setups

There are sometimes cases where the meters registered to an asset for a given utility serve more total area than this expected maximum. For example, an electric heating system provided by the landlord may serve tenant spaces as a shared service, while other electricity meters supply the smaller power and lighting in those areas. So that we don’t get coverage metrics greater than 100%, the calculation will instead take the expected area for that utility to be the total meter area, if this is greater than the GIA/NLA that would normally be expected.

To see this in action, let’s modify the asset above. Suppose the gas system is removed, and we only expect Electricity and Water Utilities. We have the same meters as before, and an additional electricity meter which heats the lettable area.

Asset Ref	Total floor area (GIA)	Net lettable area (NLA)	Common parts area (CPA)
Asset2	10,000 m²	8,000 m²	2,000 m²

Meter Ref	Utility	Procured by	Area Served (m2)	Data recorded in period?
Elec_Meter_1	Electricity	Landlord	5000	Yes
Elec_Meter_2	Electricity	Tenant	5000	No
Elec_Meter_Heating	Electricity	Landlord	8000	Yes
Water_Meter_1	Water	Landlord	5000	Yes

Then our calculations are adjusted as follows. For covered area, we now have:

Electricity	=	13,000
		+
Water	=	5,000
		= 18,000 m² floor area covered by consumption data

For electricity, the standard expected area would be the GIA, 10,000m². But since the area of the active meters declared at the asset is greater than this, we instead expect that total meter area, 18,000m², since that’s the meter setup SIERA has been provided.

Electricity	=	18,000
		+
Water	=	10,000
		= 28,000 m² meter floor area expected

Therefore, across all utilities at the asset we have:

18,000 ÷ 28,000 = 64% Data Coverage

Splitting up Landlord and Tenant Responsibility

In addition to calculating the total data coverage for the whole asset, SIERA calculates separate data coverage values for the Landlord and Tenant responsibilities at an asset. Since we’re thinking about who can collect the consumption data, we use financial responsibility to assign meters to these categories. This is indicated in SIERA by whether a meter is Landlord or Tenant-procured, i.e. who is actually receiving the invoices and paying the bills.

Let’s work through the calculation of these two subsections for the example Asset 1 above. We need to introduce one more key piece of asset data to perform this calculation, which is the asset-level management status. This is aligned with GRESB requirements – all assets are marked as Tenant-controlled if they are entirely operated by the tenant, or Landlord-controlled if there is some operational control by the owner of the building.

Asset Ref	Asset Management Status	Total floor area (GIA)	Net lettable area (NLA)	Common parts area (CPA)
Asset1	Landlord-controlled	10,000 m²	8,000 m²	2,000 m²

Meter Ref	Utility	Procured by	Area Served (m2)	Data recorded in period?
Elec_Meter_1	Electricity	Landlord	5000	Yes
Elec_Meter_2	Electricity	Tenant	5000	No
Water_Meter_1	Water	Landlord	5000	Yes

At a Landlord-controlled asset, we take the precautionary approach of assigning space to the landlord unless it’s stated to be tenant. Therefore, we first calculate the expected area for tenant supply, as the sum of NLA served for tenant-procured meters. Here this is just Elec_Meter_2, so we have 5,000m² expected tenant area. The rest of the whole building expected area (as calculated above) is assigned to the landlord. At Asset1, we were expecting 28,000m² of total meter area, so we get 23,000m² expected landlord area.

For covered area, we sum up the area served for meters that collected data in the selected period, split by Landlord/Tenant procurement.

Elec	=	5,000	Elec	=	0
		+			+
Water	=	5,000	Water	=	0
		+			+
Gas	=	0	Gas	=	0
		= 10,000 m² Landlord floor area covered by consumption data			= 0 m² Tenant floor area covered by consumption data

So, for the Landlord we have

10,000 ÷ 23,000 = 43% Data Coverage

And for the Tenant, we have

0 ÷ 5,000 = 0% Data Coverage

As we would expect, the Landlord coverage is higher than the whole building average, because we’ve separated out a Tenant coverage gap that they aren’t directly responsible for.

For a Tenant-controlled asset, the logic for expected areas would be reversed. We assume that the whole-building expected area will be the tenant’s procurement responsibility by default. If there are any Landlord procured meters (although this wouldn’t align strictly to GRESB definitions), we total up their area served, subtract it from the whole building expected area, and assign the rest as the Tenant expected area. This flips the roles of Landlord and Tenant from the case above with a Landlord-controlled asset, as you would expect.

Other Logical Considerations

The data coverage calculation also accounts for the following factors:

Inactive meters do not contribute to the expected area, or to area covered. Similarly, meters with an active from/active to window which does not overlap the reporting period are not included in the calculation.
Sub-meters do not contribute to the coverage of utilities at the asset.
Assets which were not owned in the selected reporting period are not included in coverage calculations.
Outdoor meters do not contribute towards data coverage metrics, since we typically don’t have an appropriate measure of the external area to provide an expected area for these. However, they do contribute towards data completeness.

Time-based Completeness

Data completeness shows the number of days of data that have been collected, as a proportion of the number of days of data expected in the reporting period. The “expected” days for a meter excludes days when the asset was not owned, and days when the meter was not active. Days of data are categorised as being Actual, Supplier Estimates or Calculated Estimates, based on the Origin of Data and the Actual/Estimated labels on consumption data records.

The set of meters being included in this calculation differs according to where you’re seeing the visualisation:

At fund level, all active, main meters at all assets in the fund are included by default. On the fund Data Quality page, this can be filtered to particular utilities, to assets under certain managing agents, or to meters which are landlord- or tenant-procured.
At asset level, all active, main meters at the asset are included by default. On the asset Data Quality page, this can be filtered to particular utilities, and to meters which serve certain areas of the building.

Worked Example of Completeness

The data completeness metric shows users how much data they have managed to collect in the selected reporting period, what sources the data has come from, and where there are data gaps that need filling. This calculation only includes active, main meters, and data is not expected from days where the meter was not active (indicated by the Active From and To dates) or when the asset was not owned (indicated by purchase and sale dates at asset level).

Example (12 month reporting period selected)

Full days in a year = 365
Days actual data collected = 182
Days supplier estimated data collected = 50
Days missing = 133