Article 15: What Is a Building? How the UK Counts Its Stock

A series mining the PhD thesis "London and UK Office Buildings: Investigating Energy Use and Landlord-Tenant Influences" (Azhari, 2025).

Key takeaway. There is no single answer to "what is a building" in UK data. Four agencies maintain four definitions, each optimised for a different purpose. Stock-level energy work cannot use any of them alone; it has to construct a synthesis. The Self-Contained Unit is one such synthesis, and it is the conceptual backbone of every empirical article in this series.

Figure

One physical building, seen four ways

Four UK datasets hold four definitions of the same structure. None is wrong, none is complete on its own.

VOA premise

Tax-rateable units

The Non-Domestic Rating List splits the building into rateable hereditaments (P1-P5): one occupier, one use class, one floor area each. Residential floors are not seen.

Land Registry title

Ownership polygon

One title wraps the curtilage as a single ownership envelope. The polygon labels who owns the land, not how the building subdivides.

OS Master Map footprint

Cartographic shape

One enclosed structure with one roof reads as a single TOID at ground level. Footprint is not floor area: height is needed to convert it.

EA LiDAR height

Vertical profile

Laser-scanned surface minus terrain gives a continuous height above ground (around 70 metres here). It cannot tell you the number of storeys.

The Self-Contained Unit in 3DStock is the synthesis that draws an energy boundary across all four.

The premise (VOA)

The Valuation Office Agency, an arm of HM Revenue and Customs, maintains the Non-Domestic Rating List (NDR) and the underlying Summary Valuation (SMV) data. Its job is tax assessment. To assess tax on commercial property, it must define what it is taxing. VOA does that through the concept of a hereditament, more loosely called a premise.

A premise is a unit of rateable commercial property. It typically has a single occupier, a single use class, a recorded rateable value and a recorded floor area. The floor area is what the VOA surveyors measure for tax purposes. It often differs from the Gross Internal Area used in commercial real estate transactions, and it differs again from the LiDAR-derived floor area that Ordnance Survey and the Environment Agency datasets would suggest.

Activity is encoded through the rating list use class: shop, office, factory, warehouse, leisure, education, healthcare, transport. The use classes are useful as a coarse classification and unhelpful when buildings are mixed-use. A premise classified as office might include a small canteen and a basement plant room. A premise classified as shop-and-premises might include a substantial first-floor office.

What the VOA misses is also informative. Residential premises are not in the rating list (they sit in Council Tax records, a separate system). Vacant premises are in the list but with different rateable value treatment. Some non-rated buildings (certain agricultural buildings, religious buildings, certain hospitals) are not in the list at all. For energy work, the VOA covers the commercial stock but not the picture as a whole.

The 3DStock model uses VOA SMV floor area as its primary floor-area input, but the thesis methodology had to filter cases where the SMV figure disagreed sharply with the LiDAR-derived estimate. The filter is not random; it tends to remove unusual buildings and some misclassified records. It is documented in Chapter 3.

The title (HM Land Registry)

HM Land Registry records the ownership of land in England and Wales. Its job is to define ownership, not to define buildings, but the two intersect in interesting ways.

A title is the unit of ownership. Each title has a title number, a registered proprietor, and a title plan that depicts the extent of the ownership boundary as a polygon. The INSPIRE Index Polygons dataset (originally released to comply with the EU INSPIRE directive in 2013, retained after Brexit) provides those polygons in machine-readable form. Land Registry has been gradually moving towards full digital title plans, and a substantial share of England and Wales is now mapped.

Title boundaries are not building boundaries. A title typically wraps the curtilage of a property: the building plus the land around it. Two physically connected buildings can sit on two titles. One large building can sit on one title or be divided across many. The granularity is the legal subdivision of land, not the architectural subdivision of structure.

What the Land Registry adds, for analytical purposes, is ownership. Who owns a given building is a question that VOA addresses through the rateable occupier (often the tenant) and Land Registry addresses through the registered proprietor (the freeholder or leaseholder with a long lease). For research on the organisational dimension of energy use (see Articles 3 and 6), the Land Registry is the right place to look.

The longitudinal angle matters here too. Land Registry records transactions: sales, mortgages, leases over seven years. The full history of a property is available to a researcher prepared to use the Land Registry services. For stock-level energy work, the most useful product is the simple polygon dataset (INSPIRE Index Polygons) plus the price-paid data.

The footprint (OS Master Map)

Ordnance Survey is the UK national mapping agency and the maintainer of OS Master Map, the most authoritative cartographic dataset for Great Britain. Within it, the Topography Layer is the source of building footprints.

Each topographic feature in OS Master Map carries a TOID. Buildings, parcels, paths, walls, water bodies, vegetation, roof outlines: all are TOIDs. The TOID for a building footprint is the cartographic record of the building extent at ground level.

Footprint is not floor area. A four-storey building with a 1,000 square metre footprint has 4,000 square metres of gross floor area, not 1,000. Cartographic data alone cannot tell you the difference; you need height (next section) or an external floor-area dataset (like VOA SMV).

OS Master Map identifies a footprint that is recognisably a building rather than, say, a hedge. The buildings theme has rules: a feature must be a substantially enclosed structure with a roof, of a certain minimum size. Sheds, conservatories and large garden structures sometimes qualify, sometimes do not. For most analytical purposes the OS judgment is sound.

The link to other datasets runs through TOIDs. AddressBase Premium associates each UPRN with the TOID of the building footprint it sits within. That association is what lets a researcher know that fifty UPRNs in one tower and fifty EPCs in the same tower all sit inside the same physical footprint. The address graph and the cartographic graph become a single object via TOID-UPRN bridges.

The third dimension (Environment Agency LiDAR)

The Environment Agency operates the LiDAR National Programme, a regular aerial survey using laser scanning to generate high-resolution height data across England. Wales has its own equivalent. The result is a Digital Terrain Model (the ground) and a Digital Surface Model (the top of every feature). The difference between the two, taken at the location of a building footprint, gives the building height above ground.

For 3DStock, LiDAR is the source of building height. Without it, the model would have to rely on VOA recorded number of storeys (often incomplete or inconsistent), on Ordnance Survey simplified height attribute (a coarse band), or on tax records that are not reliable. LiDAR provides a continuous, externally derived measurement that can be re-validated.

What LiDAR cannot tell you is the number of storeys. A 30 metre building could be a six-storey office tower with 5-metre floor-to-floor heights (atria, double-height lobbies, plant floors) or a ten-storey residential block with 3-metre floors. Energy analyses that depend on the number of storeys (lifts, hot-water risers, pumped systems) need an additional step, usually a regression from height to storeys with building type as a moderator.

Coverage gaps exist. LiDAR coverage is excellent across England, less consistent in rural Wales and parts of Scotland (where different agencies maintain different equivalents), and patchy in Northern Ireland. New buildings sometimes appear in the cartographic record before the next LiDAR survey captures them. For longitudinal work, the survey cadence (typically every two to five years per area) matters.

Four definitions, one building

A worked example. Take a single physical structure: a Central London building completed in 2010. Ground floor retail (a coffee shop, a sandwich bar). Floors two through fifteen office (occupied by three professional services firms). Floors sixteen through twenty-two residential (twelve flats). Twenty-three storeys total. One street address. One postcode.

VOA sees several commercial premises in the rating list: the coffee shop, the sandwich bar, and the office floors further subdivided by tenant. VOA does not see the residential floors.

Land Registry sees, depending on how the development was structured, one title (a single freehold with everything else as long leases) or several titles (separate freeholds for office, retail and residential elements, common with mixed-use developments). The polygon does not subdivide the building horizontally. It labels the ownership envelope.

OS Master Map sees one building footprint, one TOID. The building reads to the cartographer as a single enclosed structure with a single roof.

Environment Agency LiDAR sees a height surface roughly 70 metres above the ground at the building location, dropping off cleanly at the footprint edge. One height attribute attached to the footprint.

Four definitions, one building. None of them is wrong. None of them gives you a complete picture on its own. For energy analysis, you need a fifth thing.

The synthesis: the Self-Contained Unit

The Self-Contained Unit, or SCU, is the 3DStock model synthesis of the four definitions. It is defined as the smallest set of premises, footprints and meters such that no premise, footprint or meter crosses the boundary. In simple cases, an SCU is one premise inside one building with its own meters. In complex cases, an SCU spans several premises, several footprints and several meters.

The SCU concept does several things at once. It draws an energy boundary that respects metering. It collapses VOA premise-level fragmentation back to a building-relevant unit. It uses OS Master Map footprints to anchor the spatial extent. It uses Land Registry titles indirectly, through their relationship with UPRN and TOID. And it inherits LiDAR-derived height as an attribute of the footprint.

The cost of an SCU is that it is a research construct, not a regulatory one. No official agency maintains SCUs. They have to be constructed for each research project from the underlying datasets. This is non-trivial engineering work. The thesis methodology chapter is substantially given over to the construction process and the filtering rules that exclude cases where the constituent datasets disagree too much.

The benefit is that an SCU is the only unit that matches the operational reality of how a building is metered, occupied and managed. For energy work, no other available unit (premise, title, footprint) does this. For other purposes (insurance valuation, ownership audit, climate adaptation), other syntheses might be more appropriate. The principle generalises: pick the synthesis that matches your analytical question.

The 2026 National Buildings Database (DESNZ), which the author contributed to, has since applied the SCU concept at national scale. For offices alone, NBD reports 470,455 premises and 163,131 SCUs: roughly 35 SCUs for every 100 premises. The premise-to-SCU compression is the visible signature of the synthesis, and the gap between the two counts is a quick proxy for how mixed-use the stock is. Nationally, 41 per cent of office SCUs share with other non-domestic premises and 20 per cent share with domestic. The thesis Class 3 mixed-use finding from Greater London is confirmed across the country.

Why this matters beyond energy

Property due diligence asks "what am I buying?". The answer is rarely "one building". It is usually "one title, comprising five premises, with these tenants and these meters". A buyer who only looks at the Land Registry title will miss the operational fragmentation. A buyer who only looks at the VOA list will miss the ownership structure.

Decarbonisation programmes have to decide which unit they target. Heat-network feasibility runs at the SCU or premise level. EPC compliance runs at the premise level. Carbon disclosure runs at the title or portfolio level. The mismatch between these levels is a substantial source of friction in delivery.

Climate adaptation work (flood risk, overheating risk, structural risk) needs the building footprint and the LiDAR height. Insurance underwriting needs the title boundary and the building shape. Census and electoral planning need the address graph and the UPRN.

A common observation across all these applications is that the four-definition mess is not a flaw of UK data infrastructure. It is a feature. Different decisions need different units. The discipline is choosing the right unit for the question.

Practical advice for analysts

Five rules of thumb.

First, decide your unit before you decide your method. Premise, title, footprint, SCU. Different units enable different questions. The unit choice is upstream of the model choice.

Second, document the unit choice and its consequences. Filter rates, coverage gaps and the cases where the choice forced a difficult exclusion are all part of the analytical record.

Third, expect the four definitions to disagree most in complex mixed-use buildings, in long-held estates with fragmented titles, and in buildings undergoing refurbishment or change of use. Easy cases are easy. The difficulty is concentrated at the tails.

Fourth, where a third-party synthesis exists (3DStock, BBP REEB, BEIS NEED), evaluate it before building your own. Custom syntheses are expensive and the rules are easy to get wrong.

Fifth, write your unit definition down at the start of every paper. Articulate which agency definition of a building you are using, why, and what it leaves out. The four definitions are not interchangeable.

Article 4 is the umbrella piece that joins everything described in Articles 14 and 15 into one usable picture. Together the three pieces (4, 14, 15) form the methodology mini-series within the larger thirteen-piece empirical and qualitative arc.

Limitations

The article describes the documented behaviour of VOA, HM Land Registry, Ordnance Survey and Environment Agency datasets at the level relevant to stock-level energy work. It does not address the internal operational variations between local authorities, regional VOA practices or LiDAR survey cadences in detail. Northern Ireland is acknowledged but Scotland-only datasets (Registers of Scotland for titles, Ordnance Survey Scotland equivalents) are not treated. The SCU concept as implemented in 3DStock is one synthesis. Other research groups have built different syntheses for different purposes, and the article does not survey them exhaustively.

References

• Azhari, R. (2025) London and UK Office Buildings Investigating energy use and landlord/tenant influences. Doctoral thesis (Ph.D), UCL (University College London). URL: https://discovery.ucl.ac.uk/id/eprint/10204821/

• Valuation Office Agency (2012, updated periodically). Non-Domestic Rating List statistics for England and Wales.

• HM Land Registry (2013, ongoing). INSPIRE Index Polygons dataset. Available at: https://www.gov.uk/guidance/inspire-index-polygons-spatial-data

• Ordnance Survey (2023). OS Master Map Topography Layer technical specification. Available at: https://www.ordnancesurvey.co.uk/products/os-mastermap-topography-layer

• Environment Agency (ongoing). LiDAR National Programme. Available at: https://environment.data.gov.uk/dataset/f0db0249-f17b-4036-9e65-309148c97ce4

• Evans, S., Liddiard, R., and Steadman, P. (2017). 3DStock: a new kind of three-dimensional model of the building stock of England and Wales. Environment and Planning B. Available at: https://doi.org/10.1177/2399808317725161

• Steadman, P., Evans, S., Liddiard, R., Godoy-Shimizu, D., Ruyssevelt, P., and Humphrey, D. (2020). Building stock energy modelling in the UK: the 3DStock method and the London Building Stock Model. Buildings and Cities. Available at: https://doi.org/10.5334/bc.52

• Taylor, S., Lowe, R., and Steadman, P. (2014). Identifying buildings and properties in non-domestic energy datasets.

• Evans, S., Fennell, P., Humphrey, D., Liddiard, R., Oraiopoulos, A., Palmer, J., Ruyssevelt, P., Shamsi, H., Amrith, S., and Steadman, P. (2026). National Buildings Database Phase 2. Department for Energy Security and Net Zero. Available at: https://www.gov.uk/government/publications/national-buildings-database-phase-2-understanding-great-britains-buildings

Read next

• Article 4: Mapping the Stock. Inside the 3DStock Model.

• Article 14: Vertical Postcodes. How the UK Address System Handles Stacked Property.

About this series

This article is part of a fifteen-piece series adapting the 2025 PhD thesis "London and UK Office Buildings: Investigating Energy Use and Landlord-Tenant Influences" (Azhari, 2025) for a mixed academic and industry readership. The empirical findings draw on the 3DStock model of 6,038 office Self-Contained Units in Greater London with metered energy data for 2017, supplied by BEIS under a data-sharing agreement, alongside the Better Buildings Partnership Real Estate Environmental Benchmark. The qualitative findings draw on semi-structured interviews with seven major UK property organisations, conducted during the 2021 lockdown. Interviewees and their organisations are anonymised by role and organisation type. Please cite the original thesis for academic use.

Author. Rayan Azhari completed his PhD at the UCL Bartlett School of Environment, Energy and Resources in 2025, supervised by Paul Ruyssevelt and Kathryn Janda. The research was supported by the EPSRC Centre for Doctoral Training in Energy Demand (LoLo) and UK Research and Innovation through the Centre for Research into Energy Demand Solutions.

Other articles in the series. Article 1 The 30/85/89 Problem; Article 2 Why EPCs Do Not Tell You How Much Energy a Building Uses; Article 3 Eighteen Per Cent; Article 4 Mapping the Stock; Article 5 Height, Age and the Fuel Question; Article 6 The Split-Incentive Problem; Article 7 Green Leases and Service Charges; Article 8 From 38 to 73 Per Cent Energy Savings; Article 9 NABERS for Britain; Article 10 Time to Retire ECG-19; Article 11 Can London Speak for England and Wales; Article 12 The Hybrid-Work Footprint; Article 13 Why I Used Linear Regression Over Random Forest; Article 14 Vertical Postcodes; Article 15 What Is a Building?