An IoT water sensor does not control Legionella. It automates the noticing. Strip away the dashboards and the sales decks, and what these devices actually do is take a reading on their own, send it somewhere, and raise a flag when it drifts out of range — useful, but a long way from a control measure.
That distinction is the whole decision. The things that actually keep a system safe are unchanged whether or not you fit sensors: keep hot water hot, keep cold water cold, keep water moving, keep tanks and fittings clean, and keep records that show you did [1][4]. Sensors are worth the money where they close a genuine gap in how quickly and reliably you spot those measures slipping. They are an expensive screen where they don’t.
What an IoT sensor actually measures
Most water system sensors sold into UK buildings track temperature — at a sentinel outlet, on a hot flow and return, at the base of a calorifier, or in a cold water storage tank. Better setups add flow or usage data, so you can see when an outlet hasn’t run for days, which is often a more honest signal of stagnation than temperature alone. A few monitor tank level or pressure.
The common thread is automation. Instead of someone walking a route with a thermometer and a clipboard once a month, a networked device logs a reading on a schedule, pushes it over wifi or a low-power radio network to a dashboard, and compares it against thresholds you set. When a reading falls outside those limits, it alerts. That is the product, in one line: remote temperature monitoring, plus automatic record-keeping, plus alarms.
None of that changes your duties. HSE technical guidance still expects the monitoring regime — what you watch, and how often — to be set by your risk assessment, not by whatever interval a device ships with [2].
Where a sensor earns its keep — and where it doesn’t
The honest test is per-point: at this specific location, does continuous data tell me something a sensible manual check wouldn’t, and will someone act on it? The answer varies a lot across one building.
| Situation | Does a sensor earn its keep? | Why |
|---|---|---|
| Hard-to-reach or rarely-visited sentinel points (roof tanks, remote risers, an unstaffed satellite site) | Strong case | Continuous data finds temperature drift in hours, not at the next monthly visit weeks later |
| A large estate with many intermittently-used outlets | Strong case, with flow sensing | Spotting which outlets aren’t being used flags stagnation early and targets flushing where it matters |
| A busy outlet in constant daily use | Weak case | It self-flushes and gets checked anyway; a sensor here mostly produces reassuring numbers you didn’t need |
| Tank condition, scale, sediment, biofilm | No substitute | A temperature probe cannot see the inside of a tank; cleaning and inspection stay physical, hands-on tasks |
| Confirming the organism is present | No substitute | Sensors monitor conditions, not bacteria; detecting Legionella itself still needs a lab sample [3] |
The pattern is clear once you see it. Sensors are strongest where the risk hides between manual visits and the location is awkward, and weakest where a human is already there or where the question is physical rather than numeric.
The traps that turn a good idea into expensive theatre
A sensor programme fails quietly, not loudly. Four traps account for most of it.
Calibration is the first. A sensor reads precisely; that is not the same as reading accurately. Probes drift over time, and a device that confidently reports steady, in-range numbers while sitting two degrees out is worse than no sensor at all — it manufactures false confidence. Verify each sensor against a calibrated, traceable thermometer when it goes in, and on a defined schedule after that.
Placement is the second. A sensor only knows about its own point. Fit it where your risk assessment says the risk lives — the sentinel and problem points — not where the cabling happens to be easy. A perfectly calibrated probe on the wrong outlet is just tidy, useless data.
Ownership is the third, and the one that kills most systems. An alert that lands in a shared inbox no one watches is not monitoring; it is a record that you were told and did nothing. Tie thresholds to your written scheme, name the person who acts, and define the escalation before the first alarm fires. Pile up low-value alerts and people stop reading them — so tune the thresholds to mean something.
Connectivity and data ownership round it off. Plant rooms, basements and metal risers swallow radio signal, and batteries die; a sensor that has gone silent should itself raise a flag, or you are blind without knowing it. And because these readings are your compliance evidence [1], make sure you can export them in a form you can audit and keep. A proprietary dashboard you can’t get your own data out of is a liability, not an asset.
How it fits your existing controls
Treat the sensor layer as something that feeds your management chain, not something that replaces it. The risk assessment still defines what is monitored and how often [2]. The written scheme still defines what an acceptable result is and what happens when a result is out of range. The responsible person and the competent person still own the judgement calls; a green dashboard is evidence that conditions looked right at the moments they were sampled, not proof that the system is under control.
In practice the technology pairs best with a digital logbook, so the automatic readings and the human tasks — flushing, cleaning, inspections, sampling — sit in one auditable record rather than two disconnected ones. If you are weighing that wider move from paper, Paper vs digital logbooks: making the switch is the place to start; for the temperature side specifically, Remote water temperature monitoring: benefits and setup goes deeper on setup.
One caveat worth stating plainly. None of this is a substitute for a competent, site-specific risk assessment. Sensors report conditions; they do not decide what “in range” means for your building, what to do when a reading fails, or whether a given control is even needed. Every reading is only as trustworthy as the device’s last calibration. Where this overview offers a rule of thumb, treat it as a prompt to check current HSE guidance and your own assessment, not a number to copy.
What to do before you buy
You don’t need a procurement exercise to get started. You need your risk assessment and an hour.
- Open the risk assessment and list your sentinel and problem points, then mark the ones you currently check infrequently or at awkward locations. That list, not a vendor’s catalogue, is your shortlist.
- Pilot on the few hardest, highest-risk points first. Prove the alerts reach a real person and get acted on before you scale to the whole estate.
- On install day, check every sensor against a calibrated thermometer and write down the result — that is your baseline and your proof the data can be trusted.
- Decide who owns the alerts and what the escalation is before the system goes live, not after the first one is missed.
Get those four right on a handful of points and you will learn more about whether IoT water monitoring suits your site than any demo will tell you.
FAQ
Do IoT sensors replace manual temperature checks?
Not automatically. They can reduce how often someone walks a manual route, but your risk assessment sets the regime [2], and you still need periodic verification against a calibrated thermometer. Physical tasks — cleaning, descaling, inspecting tanks, taking samples — stay manual whatever you fit.
Can automatic sensor readings count as my compliance records?
Yes, if the data is complete, attributable to a known point and time, tamper-evident, and exportable in a form you can keep and audit. L8 expects records of monitoring and the management arrangements behind it [1]; a dashboard you can’t export from doesn’t meet that on its own.
Will a sensor detect Legionella in the water?
No. These devices monitor conditions that influence risk — temperature, and sometimes flow — not the bacteria themselves. Confirming whether Legionella is present still needs laboratory sampling, and that frequency follows your system and risk assessment rather than a sensor reading [3].
Sources
[1] HSE, “Legionnaires’ disease. The control of legionella bacteria in water systems - Approved Code of Practice and guidance (L8)”. https://www.hse.gov.uk/pubns/books/l8.htm [2] HSE, “Legionnaires’ disease: Technical guidance (HSG274)”. https://www.hse.gov.uk/pubns/books/hsg274.htm [3] HSE, “Testing and monitoring your water system for legionella”. https://www.hse.gov.uk/legionnaires/testing-monitoring-water-system.htm [4] HSE, “Hot and cold water systems”. https://www.hse.gov.uk/legionnaires/hot-and-cold.htm