If the sector is serious about halving leakage by 2050, it may need to rethink what success looks like – not just the leaks that are found, but the ones that are not.
The UK water sector has already made significant progress by detecting and repairing leakage. That approach remains essential. However, it is unlikely to deliver the full reduction required on its own. The next step is to better understand how leakage behaves over time, and to use data to identify change earlier rather than simply reacting to visible failure.
The value of what is not found
Across the sector, operational teams undertake inspections and routine monitoring to identify where leakage is breaking out. These activities are critical, and when leaks are found and repaired, the value is clear.
In practice, however, most inspections do not detect a leak. The majority observe no leakage signal using the methods applied. This is often interpreted as low detection efficiency.
Viewed differently, these results provide important information. They indicate that, in those locations, leakage has not increased beyond what would be expected. Repeated observations where no change is detected build confidence that the network is behaving as anticipated and that previous interventions are holding.
Low detection rates are therefore not necessarily a sign of poor performance. In many cases, they are evidence that leakage remains under control.
Why baseline matters
Leakage is not a set of isolated events. It is a continuously evolving system influenced by asset condition, pressure, demand and environmental factors.
In practice, leakage is often managed using thresholds, for example minimum night flow triggers or alarm levels set above a perceived baseline. These approaches are effective at identifying significant increases in leakage. However, they can miss more gradual changes that occur below the threshold.
From a data perspective, this is a signal detection problem. True positives identify where leakage has increased and requires intervention. Observations where no increase in leakage signal is detected (true negatives) help define what normal behaviour looks like.
Without a clear and reliable baseline, it is difficult to distinguish genuine increases in leakage from normal variability. Gradual deterioration can remain hidden until it exceeds an alarm point, at which point the system returns to reactive response.
Using data to reduce baseline leakage
Understanding baseline leakage is not the end goal. It is what enables it to be reduced.
Once baseline behaviour is established, deviation from that baseline can be detected earlier, even when changes are small. Subtle shifts in minimum night flow, pressure response or area-level behaviour can indicate emerging leakage before it becomes visible.
This creates an opportunity to intervene sooner and more effectively. Instead of focusing only on leaks that have already broken out, attention can be directed towards areas where leakage is beginning to increase.
Over time, this approach allows baseline leakage itself to be driven down, not simply held steady. Sustained leakage reduction depends not only on repairing bursts, but on systematically reducing background leakage through earlier and more targeted intervention.
Redefining success
Halving leakage will require both effective detection and a stronger understanding of how networks behave over time.
Success will not be measured solely by the volume of leakage repaired, but by how effectively baseline leakage is reduced. This depends on recognising the value of both detection and confirmation, where leakage has increased, and where it has not.
Sometimes, the most valuable result is not the leak that was found, but the many locations where no increase in leakage was detected. These observations provide the reference points needed to identify change earlier and act before leakage escalates.
True negatives do not represent absence of value. They are the foundation for understanding leakage behaviour, maintaining control of the network, and achieving sustained leakage reduction.
Artesia’s perspective
At Artesia, operational and network data is used to build decision-grade evidence that supports proactive leakage and demand management. By combining operational intelligence, network analytics and statistical modelling, behavioural baselines can be established and early deviation detected.
In many cases, the data required to do this already exists. The opportunity lies in using it to move beyond threshold-based response and towards earlier, more targeted intervention, enabling baseline leakage to be reduced over time.
