NEWS 05 November 2026

The pothole numbers that do not add up to a repaired road: Northern Ireland's defect backlog under the microscope

Published 5 November 2026 at 23:29

NI NEWS

A five-year information disclosure, a named list of the worst road sections and two official 2026 announcements reveal why headline repair totals cannot tell drivers whether the underlying maintenance backlog is shrinking.

Northern Ireland's road-repair debate is usually told in two competing numbers: how many potholes appeared and how many were filled. Official records suggest the real calculation is more difficult—and more important. In a 2024–25 road network report, the Department for Infrastructure recorded 91,715 surface defects instructed for repair and 81,483 surface defects repaired. In February 2026, the minister said 49,000 defects had been recorded in the preceding three months alone. By the end of April, a separate announcement celebrated 10,000 repairs in eight weeks. Each figure can be accurate. None, on its own, establishes that a driver facing a damaged stretch of road is now less likely to encounter it again.

The distinction matters because the department itself acknowledged in a previously published Environmental Information Regulations response that there was a backlog of instructed defects incomplete during the year because of funding and resource constraints. That disclosure also identified roads where large numbers of faults were recorded in 2024, supplied a five-year sequence of patching expenditure, and explained why a public pothole report is not the same thing as a repair instruction. Taken together, these records support a much more rigorous question than 'how many potholes were filled?': what share of the verified problems was removed permanently, how long did the rest wait, and did the stock of outstanding problems fall?

Start with the records, not the headline

The department's statistical account of 2024–25 covers Northern Ireland's 25,970 kilometres of public road. Unclassified roads accounted for 61.2% of that length, with C roads representing 18.2%, B roads 11.3%, A roads 8.9% and motorways approximately 0.4%. This network is not a single surface maintained to a uniform standard. It includes residential lanes, rural routes, busy commercial corridors and strategic roads that attract different traffic volumes and safety risks.

The same annual report says 91,715 surface defects were instructed for repair and 81,483 were repaired. The two published counts differ by 10,232, and the repairs figure is approximately 88.8% of the instructed-for-repair total. Those are straightforward comparisons. They are not a verified end-of-year backlog calculation, because repair completions in a reporting period can include instructions made earlier, while some instructions raised late in the period may be completed afterward. A genuine backlog reconciliation needs the number open at the start, additions during the year, completions, cancellations and the number open at the end.

That caveat is not a technical excuse to avoid scrutiny. It identifies exactly which data a useful performance account should publish. If the department wishes to demonstrate that the network is catching up, a year-on-year measure of outstanding repairs, separated by priority and age, would be stronger than a total of jobs completed. The annual totals show substantial activity, but cannot reveal whether that activity exceeds the rate at which the road system is deteriorating.

A publicly released list of the most defect-heavy stretches

A response issued on 10 February 2025, under reference DFI/2025-0010, listed the ten road sections with the most recorded surface defects in calendar 2024. The specificity matters. The records do not show the ten worst entire roads in Northern Ireland; they identify individual stretches with defined references and endpoints. A long road may have multiple sections, so ranking an entire route by one segment would be misleading.

At the top was the B94 Hillhead Road section between Colemans Corner Roundabout and Mill Road, associated with 180 recorded surface defects. The next was the B95 Mallusk Road section from Hightown Road to Bernice Road, with 162. Ballycorr Road between Ballyeaston Road and Church Road registered 134, while a specified Ballyrobert Road stretch and a Ballylig Road stretch each recorded 133. The response also listed sections of Derrylee Road, Glenravel Road, Aghincurk Road, Tullyah Road and Augher Point Road lower down the ranking.

Several of the highest-count sections were in the Antrim and Newtownabbey area. That is an observation about the list, not proof that this council area had the worst road condition overall. The number of defects is not divided by section length or traffic exposure, and the inspection regime can affect what gets identified. A short section with intense inspection activity and many minor faults can appear above a longer stretch with a small number of more dangerous defects. The correct use of the list is to prioritise a documented question: why did these particular segments generate so many records, and what type of treatment was carried out afterward?

What an information request revealed about the backlog

The same EIR response explains an important step that is often lost when drivers see figures for 'pothole reports'. DfI distinguished Pothole Enquiries, submitted by the public, from the Surface Defects dataset created through safety inspections and investigated complaints. A report by a motorist does not automatically become an instructed repair. Officials must examine its location, severity, risk and intervention criteria. A recorded surface defect can also be cracked, rutted, depressed or otherwise damaged without being labelled a pothole.

This gives rise to at least three distinct totals: the number of public enquiries, the number of assessed defects, and the number actually instructed for repair. They should never be treated as interchangeable. The department pointed the requester to Open Data NI's Pothole Enquiries and Surface Defects datasets, describing fields such as Client_Office_Name, Defect_Status and Approval_Status_Name. It also cautioned that some lower-priority defects are held for future maintenance planning rather than being immediately instructed for repair.

In the response, the department acknowledged that it was not possible to investigate every high-volume enquiry immediately, and explained that some faults would await the next routine inspection. It also said funding and resource constraints generated an incomplete instructed-repair backlog. This does not mean that all reported holes were neglected or that every uncompleted record represented an immediate danger. It means the public reporting chain contains several decision points and a waiting population that headline repair totals do not capture.

The money is real; the categories are different

The 2025 information disclosure supplied an especially useful cost sequence for carriageway patching. In 2019–20, DfI reported approximately £15.2 million spent on the category. The totals were about £15.6 million in 2020–21, £16.0 million in 2021–22, £17.7 million in 2022–23 and £23.9 million in 2023–24. From the first to the last year, the nominal increase was approximately 57%. It would be wrong to present that as a 57% increase in real-world repair capacity without adjusting for inflation, wages, materials and the amount of work purchased.

'Patching' itself is not a synonym for filling potholes. The department described it as small-scale work including pothole infilling and repairs to damaged areas affected by cracking or settlement. It said council-level patching expenditure was not available in the requested breakdown. That means the official figures do not support a claim that one district received a particular proportion of the pot or that a named road cost a specific sum to repair.

The wider 2024–25 road network report put total roads expenditure at £466 million, with about £136 million for structural maintenance and £102 million for new construction and improvement. These totals measure different budgets. Dividing the full £466 million by the number of patched potholes would produce a meaningless 'cost per pothole' that includes spending on work unrelated to individual road defects. Conversely, citing the patching line alone would omit larger resurfacing and structural works that may be designed to prevent recurring defects.

The winter that overwhelmed the normal reporting language

On 6 February 2026, Infrastructure Minister Liz Kimmins announced a £7.85 million Winter Recovery Road Fund following prolonged bad weather. In the accompanying statement, she said 49,000 defects had been recorded in the previous three months, a volume she described as close to half of the total for the whole previous year. The announcement followed a separate £30 million package described in early February and referred to other resurfacing schemes. The figures indicate an urgent response to an unusually damaging period, not a neat continuation of the previous financial-year series.

The number 49,000 is a recorded-defect count over a rolling three-month period. The 91,715 figure in the annual publication covers defects instructed for repair during a financial year. The two are not necessarily the same administrative category, and the periods do not align. A headline saying winter created more than half a year's 'repair backlog' by taking one figure as a proportion of the other would be unjustified. To make that claim, the records would have to use the same definition and disclose how many winter defects were already counted or subsequently repaired.

On 30 April 2026, the department announced that crews had repaired 10,000 road defects over eight weeks. That is evidence of a substantial intervention. It is not evidence that all of those repairs belonged to the 49,000 winter records, nor does the number establish their durability. The official statement did not publish a matched before-and-after list of defect IDs linking the two announcements. Without that linkage, the difference between 49,000 and 10,000 cannot be presented as the number of winter defects remaining open.

The problem with 'fixed' as a measure of success

Consider two roads with the same annual repair count. On one, a series of temporary patches repeatedly fails, and crews return to the same carriageway after every spell of rain. On the other, a planned resurfacing treatment removes the underlying deterioration and reduces the need for reactive visits. A total of repairs completed could look identical, although the outcomes for drivers and public budgets would be radically different. This is an illustration of the measurement problem, not a claim that either pattern has been measured on the named roads in the disclosure.

The most useful additional measures would be repeat defects on the same mapped section, recurrence within 90 days and within one year, time to repair by priority band, planned versus reactive intervention and the cost of repeat attendance. Data on defects that become worse while awaiting action would also matter, particularly for drivers, motorcyclists and cyclists. The existing published records tell us about observations and interventions; they do not provide a complete durability assessment in the sources examined here.

The distinction matters financially. An increase in the annual patching budget might be welcome but still leave the authority paying to revisit underlying failures. Equally, spending more on structural work may suppress the number of short-term patching jobs, which would make a fall in repair counts look like lower productivity when it could reflect better asset management. What needs to be measured is condition and risk after the intervention, not just activity during it.

Who is most exposed?

Official road-class figures show the majority of the public network consists of unclassified routes. Those roads are especially important to people whose journeys cannot easily shift to a higher-class route: rural households, deliveries, agricultural traffic and users travelling to community services. That does not establish that unclassified roads account for the majority of recorded dangerous defects. It simply makes the distribution of maintenance decisions a major public-interest question.

Drivers can have legitimate concerns about tyres, wheels and suspension, while cyclists and motorcyclists may face a different level of danger from the same damaged surface. The annual report's headline SCANNER condition measure—80% of A roads generally good and 2.4% in poor condition requiring further investigation—applies to the measured A-road class. It should not be used as though it described every residential or rural lane. And a carriageway can score acceptably in a broad condition survey while still containing an isolated safety-critical defect.

The department's own 2026 statement recognised how freeze–thaw cycles and prolonged rainfall accelerate deterioration, describing how water entering cracks can undermine underlying layers. It is an evidence-based explanation of why winter can make reactive maintenance difficult. It does not establish that every high-count stretch in the earlier disclosure was damaged by one particular storm, or that all winter defects were entirely new. The necessary evidence for those conclusions would be location-linked records over time.

Five figures the next maintenance report should publish

A more transparent annual account would publish a single reconciliation: defects open at the start, new defects verified, defects instructed for repair, completed or cancelled instructions, and defects open at the end. The figures should be reported by risk priority and age, with published definitions. The top ten named sections could then be tracked from year to year, making it possible to see whether a large investment altered their recurrence rates. DfI could also distinguish temporary make-safe works from durable repairs where its systems permit.

The second test is whether the £7.85 million winter intervention can be traced to the specific jobs it financed and the condition outcomes it produced. Third is the difference between closing a public enquiry and finishing a repair. Fourth is the availability of local expenditure data sufficient to understand trade-offs between patching and resurfacing. Fifth is publication of consistent, comparable time series so that seasonal weather claims can be tested against an unchanging definition of a defect.

These are questions raised by the documents, not answers already obtained through a new PureCar FOI. The original EIR requester secured the five-year patching costs and named road-section data; PureCar has independently compared those records with later official releases. No site inspection, engineering assessment or departmental response to these new questions is represented as completed.

The finding beneath the repair totals

The documents do not show that nobody is repairing Northern Ireland's roads. On the contrary, they record tens of thousands of completed repairs, significant annual expenditure and a large winter intervention. The harder finding is that the public numbers cannot yet demonstrate that those efforts are reducing the unresolved stock of risk. The department explicitly acknowledged an incomplete backlog in its published EIR response, and the later announcements did not supply a matched reconciliation capable of showing how much of it was eliminated.

That is why the most useful future headline would not simply count more potholes or celebrate more patches. It would state how many significant defects were outstanding, how long they had been open, which named roads were experiencing repeat failure, how much was spent preventing recurrence and whether the figures were improving on a comparable basis. Until those measures are published together, the road-repair debate will continue to have plenty of numbers but too few conclusive answers.

Records examined and analytical limits

The primary sources are DFI/2025-0010, including the February 2025 response and its road-section table; DfI's Road Network and Condition Statistics 2024–25 release; the February 2026 £7.85 million Winter Recovery Road Fund announcement; and the April 2026 10,000 repairs announcement. Counts refer to different units, time periods and administrative stages. PureCar calculated the 57% nominal increase and 88.8% repair-to-instruction ratio but does not label either an audited backlog-reduction measure. We have not downloaded and analysed the complete Open Data NI surface-defect files for this article; any full geographic ranking beyond the published ten-section table would require that separate exercise.