UK Broadband Notspots 2025: The Postcodes Still Left Behind

Ofcom's headline number for gigabit-capable broadband — 87% of UK residential premises as at July 2025, rising to 89% in the January 2026 snapshot it published in May — is a statistic about people. It is weighted by premises, so a tower block of 200 flats moves it 200 times as much as a farmhouse. That is the right way to measure a rollout. It is the wrong way to answer the question a buyer actually has, which is about one specific place: is this address still a notspot?

So we re-cut the same official data the other way. We took Ofcom's Connected Nations fixed-broadband coverage file — the July 2025 release that underpins the Connected Nations 2025 report — and calculated availability postcode by postcode, giving every postcode equal weight. Across 1,737,468 UK postcodes, 274,750 of them (15.8%) have zero gigabit-capable availability: not one premises in the postcode can order a gigabit service. And 5,877 postcodes have a majority of premises below the Universal Service Obligation — the legal minimum of 10 Mbit/s down and 1 Mbit/s up. Below is the full national picture, the twelve worst-served postcode areas, the methodology in enough detail to check our working, and an honest account of where our figures diverge from Ofcom's and why.

The national picture, measured by postcode

Four numbers describe the state of UK fixed broadband when every postcode counts once. Read these as an average across places, not across households — the difference is the whole point of the exercise, and we unpack it in the comparison with Ofcom's headlines below.

The twelve worst-served postcode areas

We ranked postcode areas by the mean share of premises below the USO, restricted to areas containing 3,000 or more postcodes so that a handful of remote postcodes cannot produce a spurious leader. Postcode areas are the letter prefix of a postcode — AB, SA, EX — and they are large: EX Exeter alone spans 22,423 postcodes across much of Devon and a slice of west Somerset.

Two things stand out. First, this is not simply a Scotland story: SA Swansea, SY Shrewsbury, EX Exeter, NR Norwich, HR Hereford, LL Llandudno and TA Taunton are all in England or Wales, and all of them sit in the top twelve. Second, the two columns measure different failures. PH Perth has the worst gigabit picture by a wide margin — 39.9% mean availability, and 52.7% of its postcodes have no gigabit option at all. TA Taunton is the mirror image: it sits second from bottom of this table on below-USO share (0.94%), yet 36.6% of its postcodes have no gigabit option, the third-highest share in the twelve, and its 58.3% mean gigabit availability is lower than eight of the ten areas ranked above it. A postcode can be comfortably above the legal minimum and still be a decade behind on the network that will define resale value in the 2030s.

The below-USO column is the one that should worry a buyer most, because it is the threshold at which the law gives you a remedy — and, as we explain below, that remedy is only free up to a point.

The other end of the table

Ranked by mean gigabit availability, the leaders are dense urban areas where a single street cabinet or fibre run reaches hundreds of premises — plus one anomaly at the top.

Hull leading the UK is not a rounding artefact — it is a consequence of history. Hull has its own incumbent network operator, KCOM, rather than BT's: Ofcom names the two as the UK's universal service providers, BT everywhere and KCOM in the Hull area. On 10 October 2019 KCOM announced completion of its £85m Lightstream rollout, saying around 200,000 homes and businesses in the region had access to 1Gbps speeds and calling Hull the first full fibre city in the UK. Nearly six years on, at the July 2025 snapshot, that decision is visible in the data as a 55-percentage-point gap between HU and PH.

Why our numbers differ from Ofcom's headlines — and why both are right

Our postcode mean for gigabit availability is 79.7%. Ofcom reports 87% — 26.4 million — of UK residential premises gigabit-capable at July 2025, rising to 89% (27.1 million) at January 2026. Our superfast mean is 94.6%; Ofcom reports superfast steady at 98%, or roughly 29.7 million, of UK residential premises. Our mean below-USO figure is 0.39%; Ofcom estimates around 44,000 premises — about 0.1% — unable to get decent broadband from a fixed line or fixed wireless access network at July 2025, falling to 39,000 at January 2026. None of these pairs contradict each other. Three things explain how they fit together:

1. Unit of analysis. Ofcom weights by premises; we weight by postcode. A postcode containing sixty flats and a postcode containing one farmhouse contribute equally to our mean. Because unserved premises cluster in small, sparse postcodes, an unweighted postcode mean is always lower than a premises-weighted percentage. That is the intended effect: we are measuring how many places are behind, not how many people.
2. Technology set. Our superfast and gigabit columns are fixed-line only, because Ofcom's coverage fields are fixed-line except where the field description says otherwise. The below-USO column is that exception: Ofcom defines it as the percentage of premises with no access to 10 Mbit/s down and 1 Mbit/s up from fixed broadband or a WISP or Mobile FWA. Fixed wireless is therefore already netted off in our 0.39%, exactly as it is in Ofcom's 44,000 — so this is not where the gap comes from. What neither figure nets off is satellite, and Ofcom's Spring 2026 update reports that cheaper Starlink plans cut the number of premises without access to affordable decent broadband to about 4,000. A house that can only reach the legal minimum via a dish still counts as a notspot here. For a buyer weighing a build quote against a satellite subscription, that distinction is meaningful rather than pedantic.
3. Availability is modelled, not measured. The postcode file records the maximum speeds operators predict they can deliver, not speeds anyone has recorded. Ofcom is explicit about this on its own coverage checker: the speeds indicated are the fastest estimated speeds predicted by the network operator(s) providing services in the area, and actual service availability at a property or speeds received may be different. Ofcom also no longer publishes a postcode-level performance file — its Connected Nations FAQs state that since 2024 there is no fixed performance postcode file, and that its analysis is robust for average download speeds down to local authority level but not for postcodes. Availability is published at postcode level; measured speed is not. Our study is therefore a study of availability, and only of availability.

Methodology

Everything above is reproducible from a free public file. Here is exactly what we did, so that a journalist, analyst or rival can check it.

• Source. Ofcom, Connected Nations fixed broadband coverage data, July 2025 release, downloaded as 202507_fixed_broadband_coverage_r01.zip from Ofcom's Data downloads 2025 page alongside the Connected Nations 2025 report of 19 November 2025. Ofcom's reference date for the snapshot is 1 July 2025.
• Which file. The postcode-level, all-premises files: 202507_fixed_postcode_coverage_r01_xx.csv, 121 files in the postcode_files folder, one per postcode area. Ofcom documents this set as containing 1,737,468 data rows, which is the denominator used throughout this piece. We did not use the residential-only postcode files (1,602,391 rows), so our figures cover residential and commercial premises together while Ofcom's headline percentages are residential-only.
• Licence. Ofcom states it provides this data on an open basis via the Open Government Licence, and its website terms of use permit reproduction free of charge provided the material is reproduced accurately, is not used in a misleading context, and is acknowledged as Ofcom copyright with the publication title specified. We have not altered any underlying values.
• Coverage. All 1,737,468 UK postcodes present in the file, including Scotland and Northern Ireland. No postcodes were excluded.
• Unit of analysis. The postcode. Area-level figures are the unweighted arithmetic mean across the postcodes within a postcode area — they are not premises-weighted. A postcode with three premises counts exactly as much as a postcode with three hundred. This is the single most important thing to understand before quoting these figures. It is also unavoidable at this level: Ofcom's field list shows that the premises-count columns (All Premises, Number of premises with Gigabit availability, Number of premises below the USO) are present at every geography except postcode, where only percentages are published. There is no premises denominator in the postcode file to weight by.
• Inclusion threshold. Postcode areas with fewer than 3,000 postcodes were excluded from the rankings, to stop small areas dominated by a few remote postcodes from topping the table on noise.
• Definitions (Ofcom's, not ours). SFBB availability (% premises) is the percentage of premises with superfast coverage, 30 Mbit/s or greater — Ofcom classes standard as below 30 Mbit/s, superfast as 30–300 Mbit/s and ultrafast as above 300 Mbit/s. Gigabit availability (% premises) is the percentage with coverage from a gigabit-capable service. % of premises below the USO is the percentage with no access to download speeds at or above 10 Mbit/s and upload speeds at or above 1 Mbit/s from fixed broadband or a WISP or Mobile FWA, including non-matched records and zero predicted speeds. The first two are fixed-line only; the third, as its definition says, already includes fixed wireless.
• Derived counts. "Postcodes with zero gigabit availability" counts rows where gigabit availability is 0%. "Postcodes where most premises are below the USO" counts rows where the below-USO percentage exceeds 50%.

What a notspot actually costs a buyer

Start with what broadband is not. It is not part of a mortgage valuation, and no standard conveyancing search covers it — the local authority, drainage, environmental and mining searches in our guide to conveyancing searches will not mention it once. Nobody in the transaction is obliged to tell you, and a seller with a 4 Mbit/s line has no reason to raise the subject. If you do not check, you will find out after completion.

What it does affect is livability and resale. Rightmove publishes a broadband speed indication on its listings — sourced, it says, from Decision Tech rather than from Ofcom, so it will not always match the Ofcom checker — which means a future buyer sees a broadband number on the advert whether or not you looked at one. On a street where every other house has fibre, a notspot narrows your buyer pool to those who do not work from home. Note too that the USO's upload floor is 1 Mbit/s — enough to exist online, not enough to run video calls and cloud backups for two people at once. Upload is the number remote workers get caught by, and it is the number listings rarely show.

And then there is the bill. Every home and business in the UK has a legal right to request a decent connection from BT (or KCOM in the Hull area) if it cannot get 10 Mbit/s down and 1 Mbit/s up. Under the Electronic Communications (Universal Service) (Broadband) Order 2018, that right runs to a cost threshold: the connection must cost no more than £3,400, or the end-user pays any costs in excess of £3,400. Ofcom puts it plainly — if building your share of the connection costs £3,400 or less you pay nothing, and if it costs more and you still want it, you pay the excess. Note what the Order does not do: it places no ceiling on that excess. Your liability is whatever the build costs above £3,400, and only the quote will tell you what that is.

• Eligibility: no access to existing decent broadband, and no publicly funded scheme due to cover the property within 12 months. BT or KCOM have 30 days to confirm eligibility and the build cost.
• Affordability trigger: you can also request a connection if the only decent service available costs more than £59.60 per month.
• Excess costs: if the build exceeds £3,400 and you want to proceed, BT or KCOM survey and quote within 60 days.
• Timescale: Ofcom says most people get a connection within 12 months, but it can take up to 24 months — a serious problem if you are moving for a remote job.

How to check a single address before you offer

1. Run the postcode through Ofcom's broadband and mobile coverage checker, which uses the same operator data as this study but at address level, and note upload as well as download.
2. Cross-check with our broadband speed checker, which reports the Ofcom fixed-line figures for the address alongside Ofcom mobile coverage — useful where 4G or 5G is the realistic fallback.
3. Compare the address against its neighbours. If the street shows fibre and this property does not, the cause is usually a long private driveway or an unadopted road, and that is a build-cost conversation, not a rollout one.
4. If the property is below the USO, contact BT (or KCOM in Hull) for an eligibility decision and a cost estimate before you exchange, not after. Only they can quote the build.
5. Fold the answer into your offer. If the fix is a quoted excess cost, it is as legitimate a negotiating point as a roof — and it belongs alongside the sold-price evidence in your ceiling price analysis.

Broadband is one of a stack of address-level facts that no viewing reveals. The same logic applies to flood risk, ground stability and mining, EPC and running costs, nearby planning applications and what the street has actually sold for — all of which our property history search pulls together for a single address in England and Wales.

Being straight about the limits: our reports pass through Ofcom's predicted availability figures. They cannot tell you the speed you will actually receive, which providers sell packages at that address, what a USO build would be quoted at, or anything at all about Scotland and Northern Ireland. For a build quote you need BT or KCOM; for a measured speed you need to stand in the house with a test.

Reuse this data

These figures are free to quote, chart, or build on. We ask two things: attribute them to HouseDossier with a link back to this page, and carry the methodological caveat with the number — these are unweighted means across postcodes, not premises-weighted national statistics, and they are not interchangeable with Ofcom's headline coverage figures. The underlying dataset is Ofcom's and should be credited as Ofcom Connected Nations 2025 fixed broadband coverage data, July 2025 release. If you want a cut we have not published — a specific postcode district, county or local authority — get in touch and we will run it.

Frequently asked questions