Retrofit feasibility studies

Retrofit feasibility, one archetype at a time.

Six building types, read from the public record. Each study takes a question an owner or programme team brings, answers it from lodged energy performance certificates, and sets out what the answer means for consents and sequencing. All six use London, BuildGrade’s first live coverage, with data as at September 2026.

Published 23 October 2026 · certified ratings only · every figure names its source and date

The studies

Pre-1930 terraced houses in London conservation areas

82.5% of certified pre-1930 terraced houses in London conservation areas are rated D or worse. Most need wall insulation, and in a conservation area the external option needs planning permission.

Certified houses: 55,344
Below band C: 82.5%

Homes in listed buildings across London

53.0% of certified homes in London's listed buildings are rated D or worse and 41.5% are single glazed. The two most frequent recommendations, wall insulation and double glazing, normally need listed building consent.

Certified homes: 45,763
Below band C: 53.0%

Post-war flats in London blocks, built 1950-1982

37.4% of certified post-war flats in London blocks are rated D or worse, the lowest share of the domestic studies, and 46.6% of the cavity-walled ones record an unfilled cavity.

Certified flats: 290,788
Below band C: 37.4%

Semi-detached houses built 1930-1949 in London

86.3% of certified 1930-1949 semi-detached houses in London are rated D or worse, the highest share of the six studies, yet 87.7% of those show a potential of C or better and few face heritage consent.

Certified houses: 140,645
Below band C: 86.3%

Pre-1930 flats in London: conversions and mansion blocks

58.2% of certified pre-1930 flats in London are rated D or worse, and they hold 32.5% of London's privately rented certificates below band C. Conversions fare worse than mansion blocks, which are more constrained by heritage.

Certified flats: 438,305
Below band C: 58.2%

Small non-domestic premises in London and MEES

Of London's certified premises of 1,000 m² or less, 3.3% are rated F or G, below the E floor in law, and 40.4% are below C. Small offices and workshops are the weakest use group.

Certified premises: 105,655
Below the E floor: 3.3%

How the archetypes compare

Six archetypes against London as a whole: latest lodged certificate per address, data month September 2026
Study	Certified	Below C	F or G
Pre-1930 terraced houses in London conservation areas	55,344	82.5%	3.7%
Homes in listed buildings across London	45,763	53.0%	2.5%
Post-war flats in London blocks, built 1950-1982	290,788	37.4%	0.8%
Semi-detached houses built 1930-1949 in London	140,645	86.3%	3.3%
Pre-1930 flats in London: conversions and mansion blocks	438,305	58.2%	1.3%
Small non-domestic premises in London and MEES	105,655	40.4%	3.3%
All certified London homes	2,526,402	49.6%	1.5%
All certified London non-domestic premises	116,189	39.1%	3.1%

Below C means rated D to G; F or G is below the E floor that applies to private lets in law today. For non-domestic premises, band C was a proposed 2027 milestone that is not being taken forward.

Method and caveats

Each study uses BuildGrade’s London dataset: 2,015,843 building units from the GLA London Building Stock Model across all 33 local authorities, matched to certificates from the MHCLG Energy Performance of Buildings register (lodged to 9 September 2026 for homes and 2 September 2026 for non-domestic premises), Historic England’s listed buildings and the GLA’s 1,089 conservation areas (both imported 12 September 2026). Certified and modelled ratings are never averaged together: 1,357,455 London units (67.3%) hold a lodged certificate, and 617,535 (30.6%) have only a modelled rating, which these studies count but do not use.

The register's age bands are 1900-1929 and 1930-1949, so a 'pre-1919' archetype and a '1945-1980' archetype could not be isolated. Studies use pre-1930 and 1950-1982 instead, and the 1930-1949 semi study includes wartime and early post-war builds.
Domestic archetype membership uses facts lodged on each address's latest certificate (property type, built form, age band), so domestic archetype counts are certified addresses by construction and unit-level certified coverage cannot be computed for them. Unit-level coverage is given for heritage-defined populations (conservation-area and listed domestic units) and for non-domestic units.
'Purpose-built', 'converted' and 'mansion block' are not recorded in the register or the stock model. The post-war flats and pre-1930 flats studies use the number of addresses in the GLA stock-model unit (2-5 vs 6 or more, from 2023 AddressBase UPRNs) as a proxy and label it as one.
Certified means a lodged register certificate (RdSAP/SAP or CEPC). GLA 2017 aggregate rows and modelled (rvEPC) ratings are excluded from every distribution. Modelled ratings appear only as separate counts of units that rely on them.
The unit of analysis for ratings is the address (UPRN): its most recently lodged certificate, with ties broken as the product does (lodged date, then record order). A stock-model 'unit' can be a whole block, so unit counts and address counts differ. 29.5% of London's latest domestic certificates are past 10 years from lodgement and may predate works since done.
Numeric domestic recommendation codes are labelled with the register's own summary text, taken from certificates BuildGrade holds with both code and text. This assumes a code means the same across register versions. Code 7's current register text reads '50 mm internal or external wall insulation', and older certificates phrase it differently. Hot-water-cylinder codes 2 and 3 were left out of the top-measure lists.
Recommendation rows are loaded for 79-97% of the latest domestic certificates and 97-99% of the latest non-domestic certificates, depending on lodgement year (some certificates carry no recommendations), so recommendation shares are lower bounds.
Cost text is reported exactly as lodged and varies with register version (PV, for example, carries '£3,500 - £5,500', '£5,000 - £8,000' and '£9,000 - £14,000'), so the most common text and its share of rows is given. The texts are not current prices. Lighting cost text varies per certificate (most common share under 15%). Non-domestic recommendations carry a payback category (SHORT, MEDIUM, LONG, OTHER), not a £ cost.
Heritage flags are spatial matches applied to the whole stock-model unit: listed means the centroid is within 5 m of an NHLE polygon or the footprint overlaps one by at least half; conservation area means the centroid is inside a GLA polygon. Every flat in a flagged block inherits the flag, so listed counts can be over or under. Listed matches: 10,294 domestic units by centroid and 5,551 by footprint.
Tenure is as recorded at lodgement and may have changed since. Certificates with unknown or blank tenure are excluded from tenure shares (tenure-known share given per study). The MEES counts are screening counts, not compliance findings, and exemptions are not modelled.
MEES status. Domestic: the E floor is law for private lets (since April 2020); band C by 2030 is confirmed policy, not yet law (government response 21 January 2026; needs legislation, and compliance is to be judged on reformed EPC metrics). Non-domestic: the E floor is law (since April 2023); the 2027 C milestone is not being taken forward, and B by 2030 was revised on 18 June 2026 to B from 2031 for buildings over 1,000 m², where cost-effective, which is not yet law. These are BuildGrade's recorded policy positions; check the current government position before relying on them.
Wall descriptions marked 'as built ... (assumed)' are RdSAP defaults where the assessor could not verify the construction (for example 79.4% of 1930-1949 semis lodged as solid brick). Treat them as lodged descriptions, not survey findings.
The potential rating on a certificate assumes every recommended measure is installed, including measures that may need planning permission or listed building consent, so 'potential C or better' is an upper bound on what is consentable.
Consent statements (permitted development, conservation-area cladding, listed building consent, flats lacking householder rights) are general planning context for England, not legal advice, and are not computed from the database. Article 4 directions vary by borough and were not queried.
London only: units outside the 33 London local authorities were not used. Other UK cities are in preparation; the Bristol and Cardiff pilots are not yet live.
Non-domestic 'small' uses certificate total floor area as a proxy for the revised proposal's 1,000 m² threshold, whose measurement basis is not yet set. One CEPC with no floor area was excluded.

What these studies do not publish

Band A on its own is merged into an 'A to B' cell in every domestic study, and A+ into 'A+ to B' for non-domestic figures, so no small count can be worked back.
Where a listed-grade count was zero (grade I or II* for 1930-1949 semis), only the total listed count is reported.
Owner and proprietor data (HM Land Registry CCOD/OCOD) is not used in these studies. No proprietor names or company numbers appear.
No addresses, UPRNs, postcodes, certificate numbers or unit references are reported. Geography is reported no finer than borough (top five boroughs per study, every cell over 3,000).

Early access

Run these questions on your own stock.

BuildGrade is in early access for design partners. Bring a portfolio, an estate or a borough and see each building’s certified rating where one is lodged, or its modelled rating labelled as such, with heritage constraints and recommended measures on one record, each with its source and date. London is live; other UK cities are in preparation.

Request early access Book an intro call