Continuous flow: how to tell a leak from genuine usage
TL;DR. The myth is that a busy site cannot tell a leak from legitimate demand, because both just look like water going out. The reality is that they look nothing alike once you measure continuous flow rather than totals. Usage stops. Leaks do not. A single leaking dual-flush loo passes 215 to 400 litres a day, every day, and on a monthly bill that is invisible.
Key takeaways
- Waterwise puts a leaking toilet at 215 to 400 litres of clean drinking water a day, on average.
- Between 5 and 8% of toilets are leaking, mostly dual flush.
- Around 400 million litres a day is estimated to leak from UK toilets.
- A monthly or quarterly total cannot separate a leak from demand, because both are just volume.
- A continuous flow trace separates them by shape: demand is stepped and returns to zero, a leak is flat and never does.
- England lost 2,690 megalitres a day to network leakage in 2023 to 2024, which is 19% of the water put into supply.
Table of Contents
On this page
Last updated: 28 August 2026
The myth: a busy building cannot tell a leak from usage
It is a reasonable belief, and it is true of the data most sites hold, because almost nobody measures continuous flow. A bill gives you a volume for a period. If the volume goes up, that could be a leak, a wetter cleaning regime, a hotter month, more occupancy, a new tenant or a chiller running harder. Nothing in a total distinguishes between them, so the argument stops there.
What makes the myth durable is that the most common source of continuous flow on a commercial site is not dramatic. It is a cistern. Waterwise records that a leaking toilet wastes between 215 and 400 litres of clean drinking water on average every day, that between 5 and 8% of toilets are leaking, mostly dual flush, and that around 400 million litres a day is estimated to leak from UK toilets.
Put one of those in a building with two hundred people in it and the extra volume is a rounding error on the invoice. Put twelve of them across an estate and it is a line item nobody can explain. Either way the continuous flow never announces itself, because a passing cistern makes no noise, causes no damage and produces no complaint.
The reality: usage stops and leaks do not
Every source of genuine demand in a building has an off state. People go home. Cleaning finishes. Irrigation runs to a timer. Plant cycles and then stops. That is the whole basis of the distinction, and it is a property of the shape of a continuous flow trace rather than its size.
Draw two days that consume exactly the same volume. The first is stepped: nothing overnight, a sharp rise at 07:00, peaks around break times, a fall through the evening, and a return to zero. The second has the same peaks but sits on a continuous flow baseline that never reaches zero. Same total, same bill, and one of them has a fault.
That baseline is the measurement. In a building that is genuinely empty overnight, continuous flow at 3am has no legitimate explanation. It is either a leak, a passing valve, a filling process nobody documented, or a tenant operating outside the hours you think they are. All four are worth knowing, and none of them shows up in a total.
How do you tell which kind of continuous flow you have?
By its stability, and by whether it responds to anything. A genuine leak through a fixed orifice is remarkably steady, because the only variable is supply pressure. Unlogged usage varies, because a person or a controller is behind it.
- Flat and unchanging overnight, week after week: a leak or a passing fitting.
- Flat overnight but different at weekends: something scheduled that nobody told you about.
- Steps up and down at odd hours: real usage on a shift pattern or a tenant with different hours.
- Rises slowly over months without a step change: a fitting degrading rather than failing.
- Disappears when a single zone is isolated: you have just located it.
Point five turns a diagnosis into a repair. Sub-metering or staged isolation narrows a whole-site continuous flow to a building, then a floor, then a riser. Without continuous flow data you cannot run that test at all, because you have nothing to watch while you close the valve.
Why does this matter beyond the bill?
Because losses on your side of the meter are counted as consumption, not as leakage, so they sit in nobody’s statistics but yours. England lost 2,690 megalitres a day to network leakage in the year to March 2024, which the Environment Agency puts at 19% of the water put into supply. That is the reported number, and it stops at the boundary.
There is a compliance dimension too, though it is worth stating precisely rather than overstating it. Regulation 4 of the Water Supply (Water Fittings) Regulations 1999 requires that every water fitting is of an appropriate quality and standard, is suitable for the circumstances in which it is used, and is installed, connected, altered, repaired or disconnected in a workmanlike manner. A cistern that has been passing for a year is not obviously meeting that description.
What do you actually need to measure it?
One meter on the incoming main, logging continuous flow rather than reading periodically. That is genuinely the whole first step, and it is worth resisting the urge to instrument everything before you know whether you have a problem.
The Clamp-on Ultrasonic Flow Meter reads bi-directional flow and total volume on live pipework across DN8 to DN100 without cutting the pipe or taking a shutdown, so a survey does not become a project. Once continuous flow shows a persistent baseline, the NOAH Multifunction Leak Sensor covers the rooms where escaping water collects, with a Leak Sensing Membrane that triggers on contact, LoRaWAN Class A and battery life up to 10 years.
Both report over encrypted cellular and LoRaWAN into the AQUAIOT Cloud, with threshold alarms on the overnight minimum rather than on the daily total, which is the setting that matters here. Alarming on volume tells you last month was expensive. Alarming on continuous flow tells you something is running right now.
Frequently asked questions
Is any overnight flow always a leak?
No, and treating it that way wastes engineer time. Plenty of sites have legitimate overnight demand: cooling plant, humidification, irrigation on a timer, a tenant working nights, an autoclave or a dishwasher on a cycle. The test is not whether there is continuous flow, it is whether it is flat, unexplained and unchanging across weeks.
Will a smart meter from the water retailer show this?
Sometimes, and it is worth asking, because half-hourly data at the point of supply will reveal continuous flow. What it will not do is tell you where in the site the flow is going, because it measures one point. Locating the source needs either sub-metering or staged isolation with something watching the trace while you do it.
How long do you need to log before the answer is reliable?
Two weeks is usually enough to establish a baseline and see whether it differs at weekends, which is the single most useful comparison. A month is better if the site has a monthly rhythm such as a cleaning deep-clean cycle or a plant changeover, because that is exactly the pattern people mistake for a leak.
Can a leaking toilet really be worth chasing?
At 215 to 400 litres a day it runs to somewhere between roughly 78,000 and 146,000 litres a year, from one fitting, and Waterwise puts the incidence at 5 to 8% of toilets. On an estate with a few hundred WCs the arithmetic stops being marginal. The repair is usually a flush valve.
Where the myth actually comes from
Not from carelessness. It comes from the fact that the only water data most sites have ever been given is a volume for a period, and a volume genuinely cannot answer the question. Told that a leak and a busy fortnight look identical, most people would agree, because without continuous flow they do.
Change the dataset and the question answers itself. Speak to an AQUAIOT engineer about metering your incoming main for a fortnight and looking at what the overnight baseline does.
By GP, Digital and IoT lead at AQUAIOT.