The commercial gigabit gap: the UK premises that cannot order a gigabit broadband connection

What these figures count, and what they do not

• A commercial premise is a property classification in Ordnance Survey AddressBase — not a trading organisation, a company or a workplace. The class can hold an empty unit, a lock-up, a storage building or a utility structure as readily as an office, a shop or a factory, and Ofcom publishes no breakdown of it. Every figure here counts premises. One organisation may occupy many premises, or none, and a premise counted here may have nobody in it.
• The national and per-nation commercial figures, including the urban/rural split, are Ofcom's own published numbers. This study credits them to Ofcom. Only the 1,011 sub-national figures, the proof of the subtraction, the unmatched bound and the side-by-side comparison are this study's work.
• Coverage is availability, not installed service or take-up. It measures whether an end user could order a service.
• Ofcom's coverage data covers broadband products only. Ofcom's own methodology says that dedicated leased lines are not included in its coverage data; other dedicated circuits are outside a broadband coverage measure too, but that is our reading and not Ofcom's wording. A commercial premise counted here as unable to order a gigabit broadband connection may hold a leased line. No correction is attempted, because per-premise leased-line availability and take-up are not published and any adjustment would be an unbounded guess.
• Unmatched premises are unclassified and are not counted as gigabit-capable. The commercial unmatched rate is several times the residential rate, so the effect is quantified rather than left as a caveat, and it is quantified twice: once assuming every unmatched premise on both sides is gigabit-capable after all, and once assuming only the commercial ones are, which is the harder case for this study's own finding.
• Ofcom's premise base admits classifications for dual-use property, places of worship, stately homes and mail-only addressees. Ofcom reports only two premise types and the split is exhaustive, so the commercial figure is the complement of the residential one and is not confined to purely commercial property. Ofcom does not publish how those records are allocated, so this is stated and not quantified.
• Rurality is published only at UK and nation level. There is no urban or rural split for any local authority or Westminster constituency, so none is given.

The number that sits in a column on its own

Ofcom publishes fixed broadband coverage for all premises and, separately, for residential premises only, at every geography it reports. It publishes the difference between the two for the UK and the four nations, and at no geography below them. That difference is the commercial premise base. Ofcom's own headline coverage figures are residential: its Spring 2026 summary workbook says of its fixed-line availability tables that all figures refer to availability at residential properties only, and a release footnote says its gigabit and full-fibre reporting focuses on residential properties. The same workbook is explicit that its take-up and its decent-broadband tables are for all premises, so the residential framing is specific to availability rather than general. Ofcom's report narrative, key findings and summary tables carry no commercial-premises gigabit coverage figure; in the surfaces checked here the commercial coverage numbers appear in the data files, and Ofcom's release footnote says its interactive report also carries coverage data, a surface not checked here.

The two files are published side by side at every geography Ofcom reports, so the subtraction is available to anyone. Before using it, this study tested it: on all 330cells across both releases where Ofcom publishes a commercial figure of its own, the all-premises count minus the residential count equals Ofcom’s commercial count to the unit, with no cell failing. The derived rows then sum back to Ofcom’s published UK totals from each geography independently, on all 18 checks run.

The national picture, which is Ofcom's own

600,791 of the UK's 2,037,917 commercial premises - 29.48% - cannot order a gigabit-capable broadband connection. For homes the figure is 11.08%, 3,373,668 of 30,455,777. The gap is 18.40 percentage points, and a commercial premise is 2.66 times as likely as a home to have no gigabit option. These national figures are Ofcom's, read straight out of its data file. Ofcom cannot match 1.551% of commercial premises and 0.213% of homes to any provider record, and does not count unmatched premises as gigabit-capable. Assume every unmatched premise on both sides is gigabit-capable after all and the gap is 17.07pp; move only the commercial side, which is the harder assumption, and it is 16.85pp. What follows below them is not.

Scroll the table sideways to see every column →

Every percentage is against that segment’s own premise base. The all-premises figure for the UK (12.23% of 32,493,694 premises) sits close to the residential one because homes outnumber commercial premises by roughly fifteen to one, which is how the commercial picture disappears from an average of the two.

Where the commercial shortfall is proportionally sharpest: towns and cities

In the urban UK 24.57% of 1,684,834 commercial premises cannot order a gigabit broadband connection, against 7.37% of 26,144,107 urban homes - a commercial premise is 3.33 times as likely to be without. In the rural UK both rates are far higher, 52.91% against 33.53%, but they are much closer together in relative terms: a ratio of 1.58. And 68.90% of the whole commercial shortfall is urban. What the data supports is a proportional claim, not a rural let-off: the commercial shortfall is sharpest in relative terms in towns and cities, and most of the shortfall is urban, while in rural areas both segments are badly served and commercial premises are still 19.38 percentage points worse off than rural homes - a wider gap in percentage points than the urban 17.20pp. Both readings are given because either one on its own would mislead. These are UK figures: they include Northern Ireland, which runs the other way.

Scroll the table sideways to see every column →

What this finding is not. Measured in percentage points rather than relative risk, the UK rural gap (19.38pp) is wider than the urban one (17.20pp), because England - and so the UK total - runs that way. In Scotland, Wales and Northern Ireland the urban gap is the wider one on that measure. The claim made here is the one that holds on both of those measures, in the UK and in each of the four nations: urban commercial premises are proportionally far worse off than urban homes, and most of the shortfall is urban. The claim that the urban gap is the widest in percentage points is not made, and neither is the claim that rural commercial premises and rural homes fare alike - they do not. Rurality is a published column in Ofcom's UK-and-nations file, beside Premise type. This cut is Ofcom's data, not a derivation of this study. What this study adds is reading the two columns together. Ofcom publishes rurality only at UK and nation level. There is no urban or rural split for any authority or constituency, so this cut cannot be mapped.

The 1,011 areas: what this study derives

Applying the same subtraction to Ofcom's local-authority and constituency files produces a commercial figure for all 361 local and unitary authorities and all 650 Westminster constituencies - the commercial picture nobody publishes at that level. The derived rows sum back to Ofcom's published UK totals exactly, from each geography independently. The gap is a consistent national and urban pattern but it is not universal: in 10 authorities and 27 constituencies, commercial premises are better served than homes.

Scroll the table sideways to see every column →

Ranked only among areas with at least 2,000 commercial premises (340 of 361 authorities, 570 of 650 constituencies). A low percentage in a tiny premise base is noise, so the floor is published rather than applied quietly.

Local and unitary authorities with the widest gap between commercial premises and homes, among areas with at least 2,000 commercial premises

Scroll the table sideways to see every column →

Westminster constituencies with the widest gap between commercial premises and homes, among areas with at least 2,000 commercial premises

Scroll the table sideways to see every column →

Ranked only among areas with at least 2,000 commercial premises (340 of 361 authorities, 570 of 650 constituencies). A low percentage in a tiny premise base is noise, so the floor is published rather than applied quietly. The count columns and the percentage columns are given together so that any row can be checked against its own denominator. All 1,011 areas, with 44 columns each, are in the published dataset.

The number that could undo this, and how far it moves it

Ofcom cannot match every premise to a provider record. Unmatched premises are unclassified and are not counted as gigabit-capable, which means they push the shortfall up. That would matter if the two segments were unmatched at the same rate, and they are not: 1.551% of commercial premises are unmatched against 0.213% of homes, an asymmetry of 7.29 times. So the bound is published rather than the caveat, and it is published twice. Assume every unmatched premise on both sides turns out to be gigabit-capable after all and the commercial shortfall falls from 29.48% to 27.93%, the residential from 11.08% to 10.86%, and the gap from 18.40pp to 17.07pp. That is the most generous reading of the coverage data, but it is not the hardest test of this finding, because it moves both segments at once. The hardest test moves only the commercial side: assume every unmatched commercial premise is gigabit-capable and leave the homes exactly as Ofcom counts them, and the gap is 16.85pp. Both are in the table. The finding survives the harder of the two assumptions as well as the more generous one. Both bear only on how unmatched premises are treated; neither tests anything else about the measure.

Scroll the table sideways to see every column →

Ofcom's own definition of its matched-premises column is “Count of all premises that could be matched to at least one fixed provider record with a non-zero predicted download speed. Unmatched premises are unclassified” (Ofcom, About this data - fixed coverage and full-fibre take-up (v2)). Unclassified premises are not counted as gigabit-capable, and they stay in the premise base as part of the denominator, so they sit inside the published shortfall. As a check on the split, the 31,608 unmatched commercial premises and 64,811 unmatched residential premises sum to 96,419, which is Ofcom’s published all-premises unclassified count exactly, with no residual. The asymmetric case. It is the smaller of the two gaps, so it is the one a sceptical reader should be given, and it is the one this study treats as the test its finding has to survive.

Six months of movement

Against the July 2025 snapshot, the commercial shortfall fell from 32.01% to 29.48%, -2.53 percentage points. The residential shortfall fell from 12.87% to 11.08%. In percentage points the gap narrowed, from 19.14pp to 18.40pp. In relative terms it widened, from 2.49 times to 2.66 times, because the residential share fell by proportionally more than the commercial share did. Both readings are given here because reporting only the first would flatter the change. These are two snapshots six months apart, not a trend. No direction is projected forward, and the next release may move either reading either way.

Scroll the table sideways to see every column →

The premise base is rebuilt from a refreshed AddressBase at each snapshot: the UK commercial premise base moves from 2,028,083 to 2,037,917. This is a change in the published measure over six months, not a tracked set of the same premises. The two releases also differ in column names, and the July 2025 'below the USO' columns are replaced in January 2026 by 'unable to receive decent broadband from fixed or FWA'; neither column is used here. The shortfall narrowed in percentage points but widened in relative terms: a commercial premise went from 2.49 times as likely as a home to be without a gigabit option, to 2.66 times as likely, because residential coverage improved from a much higher base. The per-area change columns in the published dataset are a different matter, and are joined: The 650 Westminster constituency codes, which are identical in both releases. The 361 local authorities join on code once two recoded authorities are mapped (Barnsley E08000016 to E08000038, Sheffield E08000019 to E08000039).

What a commercial premise is, and what it is not

A commercial premise is what is left when Ofcom's residential-only count is taken away from its all-premises count. It is dominated by the properties Ordnance Survey AddressBase classifies with a code beginning 'C', but Ofcom's premise base also admits dual-use property, places of worship, stately homes and mail-only addressees, and Ofcom publishes only two premise types, so those records must sit in one of the two and Ofcom does not publish which. It is a class of property, and this study is a count of premises. It is not a count of trading organisations and must not be read as one. Ofcom counts firms a different way - an address match to the Dun & Bradstreet Business Universe - and gets a different, higher coverage figure. Of that difference Ofcom says only that “Coverage remains highest for micro businesses, which are often based in residential areas and can make use of residential services”. Being based in a residential area is not the same as holding an address AddressBase classifies as residential, and Ofcom publishes no figure for how many trading premises sit in the residential count; that some do is this study's inference from Ofcom's sentence, and it is deliberately not quantified. The two measures answer different questions and neither contradicts the other.

Scroll the table sideways to see every column →

A different population counted a different way. Quoted here precisely so that the commercial premise figure is not mistaken for it. Both figures are Ofcom’s. Neither is a correction of the other.

What was measured, when, and how

Sources

• Ofcom, Connected Nations update: Spring 2026 - fixed coverage and full-fibre take-up data (January 2026 snapshot). The January 2026 national, nation, local-authority and constituency figures. File 202601_fixed_broadband_coverage_and_full_fibre_take-up-r1.zip, 33,797,696 bytes, sha256 0ddc33723e345d86…, retrieved 2026-09-29. The archive filename says r1; its top-level folder and documentation say (v2). Ofcom's correction of 07/07/2026 covers the 121 all-premises postcode files only, none of which this study uses. Ofcom names the Open Government Licence without a version number. Quoted exactly; not resolved to a version here.
• Ofcom, Connected Nations 2025 - fixed broadband coverage data (July 2025 snapshot). The six-month comparison. File 202507_fixed_broadband_coverage_r01.zip, 34,560,099 bytes, sha256 b128f5423ad7bc0d…, retrieved 2026-09-29.
• Ofcom, About this data - fixed coverage and full-fibre take-up (v2). Definitions of premise type, gigabit availability and the unmatched category; the licence wording itself; the reissue notice.
• Ofcom, Connected Nations 2025: Methodology annex. The AddressBase classification codes behind the premise base; Ofcom's statement that it normally reports residential coverage; how Ofcom counts firms differently; the leased-line exclusion; the third-party attributions Ofcom's SME source carries. This document carries no licence statement of its own. The sentence quoted for the data releases appears only in Ofcom's 'About this data' documents, so it is not repeated here. Ofcom's site terms of use, and the attribution condition quoted beside this entry, apply.
• Ofcom, Connected Nations UK report 2025. Ofcom's published SME coverage figures (Table 2.3, page 10), quoted so the two measures are not confused; the leased-line exclusion. This document carries no licence statement of its own. The sentence quoted for the data releases appears only in Ofcom's 'About this data' documents, so it is not repeated here. Ofcom's site terms of use, and the attribution condition quoted beside this entry, apply.
• Ofcom, Connected Nations Spring 2026 update tables. The workbook's statement, made of its fixed-line availability tables, that all figures refer to availability at residential properties only - and its separate statements that its take-up and decent-broadband tables are for all premises. This document carries no licence statement of its own. The sentence quoted for the data releases appears only in Ofcom's 'About this data' documents, so it is not repeated here. Ofcom's site terms of use, and the attribution condition quoted beside this entry, apply.

Where sources disagree

• Ofcom's commercial-premise coverage against Ofcom's own SME coverage. Ofcom, fixed coverage data files, commercial premise classification: 68.0% of commercial premises gigabit-capable at July 2025; Ofcom, Connected Nations UK report 2025, Tables 2.3 and 2.4, page 10: 84% of SMEs gigabit-capable at July 2025. Both are Ofcom's and both are correct. They count different populations by different methods: a property classification in AddressBase against an address match to a commercial register. This study reports the premise figure and names the other so the two are not confused. No winner is picked.
• The upper-bound gap when unmatched premises are assumed gigabit-capable. one data-collection stream of this study, moving only the commercial unmatched premises: 16.85pp; a second, independently written stream, moving both segments: 17.07pp. Recomputed here from Ofcom's published counts: (600,791 - 31,608) / 2,037,917 = 27.93%, and (3,373,668 - 64,811) / 30,455,777 = 10.86%, giving 17.07pp. Neither stream was wrong. They applied different assumptions: the higher figure moves both segments, the lower moves only the commercial one and leaves the homes as Ofcom counts them. The lower is the harder case for this study's finding, so rather than pick a winner the page now publishes both, each with its assumption named, and says which one binds.
• The landing-page URLs recorded in the study brief. the study brief: /research-and-data/multi-sector-research/infrastructure-research/connected-nations-spring-2026; Ofcom's live site, fetched 2026-09-29: /phones-and-broadband/coverage-and-speeds/connected-nations-update-spring-2026. Ofcom restructured its site; the brief's URLs redirect and then 404. The working URLs are cited throughout and the archived landing pages are held as evidence. The data files themselves are unaffected and match their expected byte counts and hashes.
• Ofcom's own documentation against Ofcom's own data. Ofcom, About this data (v2): the constituency name column is 'parl_const_name'; 202601_fixed_pcon_coverage_r1.csv: the column is actually 'parliamentary_constituency_name'. The file wins, because the file is what anyone parses. Reported so that a reader reproducing this work is not caught by it.

Prior work, and what this adds

• Ofcom: Publishes the commercial premise counts and gigabit counts for the UK and the four nations, split urban and rural, as columns in its fixed coverage data file. What this study adds: We derive the same figure for 361 local authorities and 650 Westminster constituencies, which Ofcom does not publish at any geography below nation level, and we prove the subtraction before using it.
• Ofcom: Publishes SME gigabit and full-fibre coverage by nation, urban/rural and size band, using an address match to the Dun & Bradstreet Business Universe. What this study adds: A different population and a different method, giving a different and higher figure. We report the premise measure and name Ofcom's other measure beside it rather than blending them.
• Ofcom: Its headline gigabit and full-fibre coverage reporting, which is residential only and says so. What this study adds: We place the commercial figure next to the residential one, which is the comparison the separate publication of the two files makes possible, and which we found no published example of in the surfaces we checked.
• Office for National Statistics, Explore local statistics: Publishes a 'Gigabit capable broadband' indicator - 'the percentage of premises that have coverage from a gigabit-capable service (download speeds of 1,000 Mbps or more)' - for lower tier and unitary authorities, from Ofcom's own January 2026 snapshot, published 13 May 2026 under the Open Government Licence v3.0 — the ONS's own licence statement, not Ofcom's, which names no version. It is the same snapshot and one of the two geographies used here. What this study adds: It is an all-premises figure and does not distinguish residential from commercial, which is the whole of what this study adds. We also cover Westminster constituencies, which it does not, and publish the residential figure beside the commercial one so the two can be compared. Retrieved 29 September 2026 and archived with this study.
• Others: Two third-party sites that publish constituency-level broadband material were checked. Both returned HTTP 403 to an identifying client, which was treated as a signal not to fetch, and no further request was made. What this study adds: Their contents are unverified and are therefore not characterised here, in either direction. This is a limit of the prior-art search, recorded rather than papered over. No figure in this study depends on them.

Ofcom publishes the commercial premise figures for the UK and the four nations. We found no published commercial-premises coverage figure for any local authority or Westminster constituency in the surfaces we checked, because Ofcom publishes no commercial file and no premise-type column at those levels. What was checked: Ofcom's own downloadable file inventory and the archive member lists, which at every sub-national level publish exactly two files, all premises and residential only; and the nearest third-party publication, the ONS Explore local statistics gigabit indicator, which carries the same Ofcom snapshot at lower-tier authority level for all premises with no premise-type split. What was not checked: Ofcom's interactive report, which its release footnote says also carries all-premises coverage data, and two third-party sites that returned HTTP 403 and were left alone. Not described as a first. The national figure is Ofcom's and is credited as such. Nothing here is claimed about the surfaces that were not checked. Refresh from the new release or date the page clearly to the January 2026 snapshot. The dataset and README carry the same instruction. The next Connected Nations release is expected around November 2026.

Source: Ofcom, Connected Nations update: Spring 2026, fixed coverage and full-fibre take-up data (January 2026 snapshot), and Connected Nations 2025, fixed broadband coverage data (July 2025 snapshot). Ofcom copyright. Ofcom provides this data on an open basis via the Open Government Licence. Downloads: CSV, 1,011 rows · README and attribution · JSON. CSV sha256 9b567123f27fa661…, README sha256 7e899521336fb614….

Questions

If the connection is the constraint

This study counts premises, not organisations. Separately: where an organisation's own site cannot order a gigabit broadband service, the decisions that follow are about how much compute, storage and backup sits on site rather than at the other end of the link. These three tools take the connection as an input rather than an assumption.

Studies built the same way, from published official data: large-site electricity by local authority · where UK government cloud data is declared to go · the UK public sector IT contract expiry cliff · all Servnet research