By Alistair Brown, CEO of LumenStream, and Matt Clark, Technical Lead at LumenStream.
Last updated: 17 October 2026
In September, Matt, our Technical Lead, stood under a mixer at a food production site and asked the site manager what happens if the lights fail while someone is up there.
The site had heavy machinery, a skilled operator working at height to tip ingredients in, and no emergency lighting over that spot at all. The manager's answer: "They have to use their mobile phone, and they have to use their lights on their mobile phone."
Matt's word for his own reaction was 'dumbstruck'. That site, he says, is like a lot of them: the lights come on in the morning and go off at night, and that is considered to be fine.
Which is the exact problem this article is about.
Your lighting is compliant when three things are true at once:
- The light levels suit the work in each room.
- The emergency lighting covers the building as it is used today, and it is tested on schedule.
- And you can show an inspector or insurer the records that prove both.
"The lights work" is none of the three.

If you are the health and safety manager, the responsible person for fire safety, or the facilities manager who gets the call when either of them has a question, this guide is for you.
Check 1: Are the light levels right for the work you do in each room?
The first check is whether each room has enough light for what people do in it, measured, not assumed.
What does UK law require from workplace lighting?
The law requires "suitable and sufficient" lighting in every workplace, and it leaves the numbers to guidance and standards.
That wording comes from regulation 8 of the Workplace (Health, Safety and Welfare) Regulations 1992, which applies in England, Wales and Scotland. The same regulation adds a second duty most sites have never read. Where people at work are "specially exposed to danger in the event of failure of artificial lighting", that room needs suitable and sufficient emergency lighting too.
In Matt's example, a mixer is that room.
The regulation gives no specific lux figure, but two documents fill that gap:
- The first is HSE's own guidance, HSG38 Lighting at work (second edition, 1997, still HSE's current guide). It gives average figures by the detail the work needs: 20 lux for corridors, 50 for loading bays, 100 for kitchens and assembly of large components, 200 for offices and sheet metal work, and 500 for drawing offices and electronic assembly.
- The second is the design standard UK lighting designers work to, BS EN 12464-1:2021. It sets higher numbers for the same spaces, because it is written for comfort and productivity rather than as a safety minimum. It asks for around 300 lux in general manufacturing areas and 500 lux for office tasks.
So when a contractor quotes you 500 lux for an office and HSE's guide says 200, neither is wrong: the guide gives HSE's average for the work (with 100 as the measured minimum), and the standard gives the level a new installation is designed to.
What lighting level does my site need?
Your site needs the level that matches the task in each room, and the only way to know whether you have it is to measure it with a lux meter at the working plane.
Our guide for facilities managers carries the full table by space type, and HSE's lighting guidance puts a process control room at 300 lux and close work such as studying an engineering drawing at 750. The finer the detail the work involves, the more light it needs, and the number that counts is measured where the work happens, not at the ceiling.
Two things a meter reading alone will miss.
- Uniformity is the first. A room can hit the average and still have a dark corner where a fitting failed and nobody noticed because the room "looked fine".
- Glare is the second: a bare high-bay in the line of sight can meet the lux figure and still stop an operator reading a label.
For example, at Siemens, one of our UK customers, we replaced 2,426 fittings and the new design lifted light levels to over 500 lux, above the UK standard for manufacturing. The number to look for on your own site is the one taken after an install, not the one on the design drawing.
Why does a site that passes on average still fail?
A site passes on average and fails in the one room that matters because the average is built from the corridors and the offices, and the risk is on the production floor.
This is why Matt breaks a site down room by room rather than scoring it as a whole. Several rooms come back fine, and the one that does not can be the room with the heaviest machinery and the most people in it, which a site-wide average hides. Often the lighting was designed for a building that no longer exists.
In Matt's words, a storage area becomes a production area, and nobody redraws the lighting when the use changes.
Measure room by room, and record the rooms that fail with their readings. That list is the first page of your evidence file.
Which brings us to the second check:
Not having enough emergency lighting which is, in Matt's words, a typical reason a site fails.
Check 2: Does your emergency lighting cover the building as it is used today, and is it tested?
The second check is whether every escape route and every high-risk work area has emergency lighting that will come on when the mains fail, and whether someone is testing it on the schedule the code sets.
Emergency lighting is a legal duty, not a recommendation, wherever it is needed to keep people safe. In England and Wales the duty sits in article 14 of the Regulatory Reform (Fire Safety) Order 2005: "emergency routes and exits requiring illumination must be provided with emergency lighting of adequate intensity in the case of failure of their normal lighting." Scotland's regulations and Northern Ireland's use the same words.
The workplace regulation above adds the danger-area duty on top.
Three ways Matt sees sites fail this check.
- There is no emergency lighting at all over the place where someone would be most exposed if the lights went out. The mixer is one example.
- The schedule says there is, and the ceiling disagrees. On one of the sites we surveyed, the schedule was in the folder and the key switch was on the wall. Twenty of the fittings on that schedule had been swapped for standard fittings years earlier, because the emergency versions cost a little more, and the swap saved a couple of hundred pounds.
- There is one emergency fitting for the whole building. Take a site that for argument's sake has run that way for 10 years and says "we've got emergency lights". Matt calls that the more difficult conversation. The duty is on the building as it is used today, not on the day the fitting went in.
How often must emergency lighting be tested, and by whom?
Emergency lighting is tested monthly by whoever is responsible on site, and once a year by a competent person, with every test written down.
The government's own fire safety guidance for the responsible person sets the regime. Check the units are in good repair, daily where practicable. Test every system monthly, "at an appropriate time when, following the test, they will not be immediately required". Have a competent person test and maintain the system once a year. Keep a record of every test.
The annual test is the full-duration one, three hours for most buildings under the code.
The Fire Protection Association's guide (2023) explains how it differs from the monthly "flick" test. Batteries typically take 24 hours to recharge afterwards, so the annual test is planned for a time when the building can stay empty, or other cover is arranged, until they are back.
To give you the scale of it: for one 50-fitting factory we priced, the eleven monthly tests and the annual three-hour test came to £2,086.50 a year.
That is the cost of the testing itself.
On a recent visit, the person Matt spoke to put the whole thing down to cost. His reaction was straightforward: at what cost is someone's health, or their life?
What changed in October 2025?
The emergency lighting code of practice was rewritten last October, and the change Matt has yet to find on any site he visits is a five-year lux record.
BS 5266-1:2025 was published on 29 October 2025 and came into effect on 31 October, replacing the 2016 edition. A new clause asks for photometric verification of the system: a measured check that the emergency lighting reaches the required lux level, when the system is installed and every five years after that. NICEIC's summary of the changes lists it as clause 11.2.
The levels are small numbers:
- 1 lux along an escape route
- 0.5 lux in an open area
- And, where a high-risk task has to be made safe before someone leaves, a tenth of the normal level and never less than 15 lux
This is a code of practice, not a statute. Matt sat a refresher course in August and found rules from the last six months he did not know; he has not yet visited a site that complies with all the latest regulations.
Which is the problem with the third check: a site can have done the work and still have nothing to show for it.
Check 3: Could you show an inspector or insurer the evidence tomorrow?
The third check is whether the proof exists on paper or in a system, in a form someone who has never visited your site could read.
What does an inspector ask to see?
An inspector asks for written evidence as well as looking at the fittings.
Warwickshire Fire and Rescue's guidance lists what that evidence is:
- Written declarations of compliance available on site
- A completion certificate from the installer
- Photometric evidence from the system designer
- And "a log of all system tests and results"
Under the 2025 code, add the five-year lux record. For workplace lighting, it means your risk assessment and the readings behind it. Then the records get checked against the ceiling, which is how Matt found the 20 swapped fittings.
This is not a rare event.
In the year to March 2026, fire and rescue authorities in England completed 50,195 fire safety audits, and 60% came back satisfactory. The most cited reason for failing was a breach of article 14, the emergency routes and exits article that carries the emergency lighting duty, up 9.3% on the year before. Formal notices rose 11% to 3,355 (MHCLG, published September 2026).
The release does not separate emergency lighting from the rest of article 14. It does show that the article your emergency lighting sits under is the one audits fail on most.
Before Matt looks at a single fitting he asks the site contact three questions: what the current testing procedure is, whether he can see the emergency lighting logbook, and whether the site has had any emergency lighting issues or concerns. In his experience the answers to those three settle most of the audit before anyone opens a ladder. If you can't answer all three today, that is the gap to close first.
Will our insurer ask about lighting?
Your insurer may ask for your emergency lighting test records, and you should check your policy rather than wait to find out.
We looked for a published insurer requirement on emergency lighting and did not find one. Policy wording varies and most of it is not public. What we can tell you is what Matt has seen over the last two to three years: more insurers asking for proof of an ongoing test procedure and an up-to-date certificate, the way a car insurer asks you to prove your no-claims history.
In his words, they are always looking for a way of not paying. Ask your broker in writing what your policy requires, and file the answer with the logbook.
Does switching to LED make us compliant?
An LED upgrade fixes the light levels and lets you design emergency cover into the new layout.
It does not do your testing or keep your records for you.
A proper redesign starts from the task in each room, so a site that failed Check 1 on light levels passes it after the install, as the Siemens and Fedrigoni figures show, and uniformity and glare are dealt with at the design stage. Emergency fittings go into the same layout, over the escape routes and danger areas as the building is used now, which closes the "designed for a building that no longer exists" problem.
What stays with you is Check 2's schedule and Check 3's paperwork.
Self-testing emergency fittings, in Matt's words, take out the weekly walk-round and turn the monthly test into pressing a button. You still need a person for the annual test, and the record of every test still has to exist.
If someone tells you an LED upgrade sorts compliance on its own, ask which of the three checks they mean.
In Matt's experience a lot of sites rely on the lighting manufacturer's rep to have sorted it, and that is where the belief comes from.









