210,972 streetlight faults: the NI decline that still needs explaining
Northern Ireland recorded 210,972 repaired streetlighting faults in six calendar years. Reports more than halved overall, but several districts followed strikingly different paths and the published data does not explain why.
The fall is substantial — and uneven
Environmental Information Regulations response DFI/2025-0117, issued on 18 April 2025, contains a council-by-council record of streetlighting faults repaired by DfI Roads’ own teams and external contractors. Summing its six annual Northern Ireland totals gives 210,972 faults for 2019–2024. The annual figure fell from 48,325 in 2019 to 21,499 in 2024, a reduction of 26,826, or 55.5%. These are records of faults, not a headcount of broken streetlights: the same asset could potentially generate more than one fault over a long period.
| Calendar year | Faults in response | Change on preceding year |
|---|---|---|
| 2019 | 48,325 | — |
| 2020 | 46,297 | −4.2% |
| 2021 | 37,839 | −18.3% |
| 2022 | 31,752 | −16.1% |
| 2023 | 25,260 | −20.4% |
| 2024 | 21,499 | −14.9% |
The most misleading interpretation would be to say this proves Northern Ireland’s streetlights are 55.5% more reliable. The table does not provide operating hours, asset ages, the number of lights in each district, whether reporting methods changed, or whether repairs to more than one defect were grouped under a single report. It combines outage and non-outage faults. Reliability is a plausible hypothesis, not an established result.
A district gap hidden behind the Northern Ireland total
Ards and North Down recorded 5,558 faults in 2019 and 4,942 in 2024 — a fall of only 616, or 11.1%. Belfast declined from 8,114 to 3,045 over the same dates, a drop of 5,069 or 62.5%. Fermanagh and Omagh fell from 2,365 to 531, down 77.5%, while the Armagh, Banbridge and Craigavon total went from 3,378 to 2,378, down 29.6%. These are different trajectories under the same departmental reporting system.
| Council district | 2019 | 2024 | Change |
|---|---|---|---|
| Ards and North Down | 5,558 | 4,942 | −11.1% |
| Belfast | 8,114 | 3,045 | −62.5% |
| Fermanagh and Omagh | 2,365 | 531 | −77.5% |
| Armagh, Banbridge and Craigavon | 3,378 | 2,378 | −29.6% |
| Derry and Strabane | 5,244 | 2,036 | −61.2% |
If this were a proper league table, every district would first need an asset denominator: how many DfI lights are actually maintained within the council boundary? A town-centre district with many lamps can have more incidents than a rural district even with better condition per asset. Fault severity and repeat visits also matter. The raw table is useful for asking questions, not labelling an area the worst maintained.
The fact that Ards and North Down was relatively flat against a sharply falling NI average deserves investigation. Possible explanations include the condition and age of the local network, asset replacement schedules, repair reporting processes and contractor arrangements. None of those possible causes is established by this response, so PureCar is not attributing fault to any council or contractor.
The £15.1m cost figure is from different years
The same response provides repair costs for six financial years, 2018/19 to 2023/24: £2.578m, £2.052m, £3.205m, £3.200m, £1.994m and £2.100m. The sum is £15.129m. It is tempting to divide that total by the six-calendar-year fault count, but doing so would create a false cost-per-fault metric because the periods do not align. The departmental disclosure explicitly says it does not hold repair expenditure by council area in the form requested.
Costs can also change because of work complexity, contract prices, component costs, labour and emergencies. A year with fewer simple bulb faults but more expensive cable or column repairs might cost more. The table cannot separate routine repairs from those drivers. In other words, falling fault counts and the cost line tell different stories, not a single proven efficiency result.
For perspective, DfI said it was responsible for around 290,000 streetlights. If every one of the 21,499 records in 2024 represented a different light — something the data does not establish — that would equal approximately 7.4 records per 100 lamps. It is not a 7.4% failure probability and not proof that 7.4% of lights were simultaneously out.
Repair targets are not evidence of every repair
DfI’s disclosure says streetlighting faults are generally repaired within five working days after an initial report and that at least 95% of lights are working at any given time. These are statements of service practice and network availability, not a published table showing that every reported fault met a five-day deadline. Separate DfI guidance distinguishes emergency, urgent and non-emergency defects; exposed wiring and structural hazards are not equivalent to a lamp that has stopped illuminating.
For motorists and pedestrians, the distinction matters. A dark junction, a school crossing and an isolated residential lamp do not necessarily present the same immediate hazard. Nor can this dataset link any particular collision to a missing or faulty light. A responsible road-safety investigation would need geographic and time-specific evidence before making such a claim.
The official nidirect service allows people to report a public streetlight fault and track its progress. Residents should provide the pole identifier if available, the street, a precise location and the nature of the hazard, rather than assuming that a previous report by another person has been logged correctly.
What a more revealing public dataset would contain
An annual district return would be much more informative if it published the number of maintained assets, new reports, repeat faults, average repair time, the share exceeding the target, the severity category and the number of reports still open at year end. It could also show the proportion of the network converted to LED technology; nidirect identifies LEDs as the preferred option for new and replacement schemes, but the disclosure does not demonstrate whether conversion caused the trend.
That is PureCar’s central finding: the published record is sufficient to identify a major six-year decline and notable geographic exceptions, yet insufficient to attribute success or failure confidently. The headline decline should invite more scrutiny, not less. If the dataset is used to decide where lighting money goes, its missing denominators are as consequential as the totals themselves.
How to test competing explanations without manufacturing one
Suppose LED replacements reduce the frequency of some lighting faults. A decline in repairs might then be an operational success. But suppose the number of reports fell because faults were bundled into larger work orders, inspection schedules changed, or members of the public reported fewer issues. The same aggregate chart might appear even when the underlying network experience was different. The current disclosure does not contain the fields necessary to separate those scenarios.
A credible follow-up would ask for the number of lamp columns in each district for each year, records separating electrical outages from other defects, how duplicate reports are handled and the number of repeat failures at the same asset. It should also examine the proportion of repairs performed within the stated five-working-day period and report any variation by severity. Without those controls, total counts measure administrative events more confidently than they measure the service experienced on a dark road.
There is a further timing issue in the published material. Its repair-cost table begins in April 2018 and ends in March 2024, while the fault-count table begins in January 2019 and ends in December 2024. A credible cost-per-fault trend would align individual financial periods to a matching fault dataset and establish what kinds of work are included in expenditure. These basic choices determine whether a performance story is meaningful or merely a division of two available numbers.
Sources and methodology
PureCar examined the original published government response and cross-checked the interpretations against the official guidance linked below. Calculations are derived from the published figures; they are not new data supplied privately to PureCar. The response dates and coverage are stated in the article because historical disclosures must not be represented as newly released information.