The six jobs
These are every investigation in our records over the period that was triggered by a meter reading or a notice from a water company, with the reported loss converted to a common rate where the record gives enough to do it:
| Reported loss | Roughly | What it turned out to be |
|---|
| 168 litres a week, per a water company notice | 1 litre an hour | No leak in the property at all. The supply was shared with the flat above and the discharge was theirs |
| Around 100 litres a day, per a water company notice | 4 litres an hour | Two WC cisterns passing, one badly, plus a dripping bath tap |
| A bill that had roughly doubled | Not quantified | A loft tank ball valve and a WC filling valve, both passing continuously |
| 25 litres an hour on the meter | 25 litres an hour | A joint between old imperial-sized plastic underground pipe and a stopcock, below a floor |
| Around 2,000 litres a day | 83 litres an hour | The buried incoming supply pipe, roughly a metre down. Confirmed by elimination, not pinpointed |
| 7 litres a minute on the meter | 420 litres an hour | A hole in a lead supply pipe under a floor |
Sorted by size, the three smallest were fittings or nothing, and the three largest were all pipes, two of them buried or underground. That is a clean pattern and six jobs is nowhere near enough to make it a rule. Treat it as a hypothesis worth holding lightly, not a diagnostic.
Why small losses tend to be a toilet
There is a physical reason the pattern points this way, which is why it is worth mentioning at all despite the sample size.
A WC filling valve or a siphon that does not seat lets water trickle continuously into the pan, where it goes straight down the waste and leaves no mark. The same goes for a ball valve in a loft tank, which discharges to an overflow, and for a tap seat that has worn. All of these produce a steady, unremarkable rate and no damage at all, which is exactly why they are found by a meter rather than by a stain.
A failed pressurised pipe behaves differently. It is escaping into a structure rather than into a drain, so the same volume causes visible harm, and the rates observed here were much higher. The 420-litre-an-hour job had a reception room floor that had visibly sunk.
The practical consequence: if your loss is measured in litres per hour rather than litres per minute, and nothing in the building is damp, check every toilet, every tank and every tap before anyone suggests lifting a floor. The full checklist is here.
The visible drip that is not the answer
In two of the six, the engineer found a dripping tap within minutes of arriving and recorded, in both cases, that it was nowhere near enough to explain the loss. One of those properties was losing around 420 litres an hour.
This is the single most common way a metered-loss job goes wrong. A drip is found, it is fixed, the invoice is settled, and the meter keeps running. If the defect you have found cannot plausibly produce the volume the meter is reporting, it is not the whole answer. Do the arithmetic before you accept it: a fast drip is measured in litres a day, not litres a minute.
The one where the property had no leak
A water company notice reported an extra 168 litres a week, about a litre an hour, and suggested a slowly leaking toilet or tap as the likely cause. Three separate visits followed. The engineer did a full visual inspection, pressure tested hot and cold, ran thermal imaging across the accessible floors, put an inspection camera under the bathroom floorboards and ran tracer gas on the cold system. Every test was negative.
What the investigation did find was that the external stopcock served both that property and the flat above it: closing the internal isolation point cut the upstairs supply too. An external inspection then found a continuous drip into the hopper serving the upstairs flat, from their bath and basin waste.
Two lessons. The shared supply also meant tracer gas introduced into the system could migrate into the neighbouring property, which compromises the test itself. And if you are on a shared supply, your meter can report someone else's water. Who pays in that situation is a separate question with a separate answer.
The one that could not be pinpointed
A large building was losing roughly 2,000 litres a day. The method here is worth describing because it is the cheapest useful test there is.
The internal stopcock was closed while the external supply stayed open, and the meter was watched. It kept registering flow. Every other internal stopcock that could be found was then closed in turn, and the meter still ran. At that point the loss is proven to be downstream of the meter and upstream of the building's internal plumbing, which is to say in the buried service pipe, and no amount of looking inside the building will find it.
Tracer gas was introduced and the route walked with detection equipment. Nothing surfaced. The pipe was about a metre down beneath a hard surface, and at that depth both the water and the gas disperse into the ground rather than reaching the top. The investigation ended with the loss confined to a stretch of buried pipe, and the recommendation being to obtain the original pipe route from the water company or the building records and excavate to it rather than dig speculative trial holes.
That is an unsatisfying outcome, and it is an honest one. A report that says "the loss is in this thirty metres of buried pipe and here is why" is worth considerably more than one that guesses a spot.
Before you book anything
- Take two meter readings an hour apart with every outlet closed. No taps, no washing machine, no dishwasher, nobody flushing. A moving meter is proof; a suspicion is not. The method is here.
- Close your internal stopcock and read the meter again. If it still moves, the loss is between the meter and your stopcock, which usually means the underground supply pipe. If it stops, the loss is inside.
- Put a sheet of toilet paper against the back of each pan, above the waterline. If it wets, that cistern is passing. This found the fault in two of these six jobs and costs nothing.
- Check the loft tank and any overflow pipe outside. A discharging overflow is a passing ball valve and it never shows up indoors.
- Keep the water company's letter. It is dated, it states the volume, and it is the strongest evidence you have when you later apply for an allowance.
Thames Water publishes what counts as a leak on your side of the boundary and how its leak allowance works, and the Consumer Council for Water sets out what you can do when you suspect a leak on your own supply. Both are worth reading before you commit to a detection visit.
What it costs, and when not to bother
Detection is charged at £150 per hour, agreed before we attend, and under no find, no fee you do not pay for the time spent looking if we do not find it. The total depends on hours, so a small internal loss that turns out to be a cistern is a very different bill from a buried supply pipe under a hard surface.
Do not book detection if you have not yet done the stopcock test. It takes ten minutes and it determines whether you need an internal investigation or a supply pipe job, which are different visits with different equipment. And if your only evidence is a bill that looks high, read the piece on telling a leak from a change in usage first. A new occupant, a garden through a dry summer or a faulty meter all look like a leak on a statement.
Limits
Six jobs. Six is not a sample, it is an anecdote with a table round it, and we have presented it as one. The losses were reported by customers and water companies rather than measured by us in a standardised way, the "bill had doubled" job is not quantified at all, and the properties are self-selected by who called us. The wider method, universe and exclusions are set out in the main dataset post.
