Our leakage performance

Interesting fact for April 2020

Finding and fixing leaks is one of our biggest priorities, and we're doing lots to reduce our leakage every day.

However, did you know that about two-thirds of the leaks we fix on our pipes are hidden underground, so we spend £1 million a day and have teams of people working day and night to uncover them. This helps us repair almost 1,400 leaks on average every week.

About two-thirds of repairs on our pipes are non-visible leaks.

We hope you find this interesting – and please remember to help us save water by using it wisely, whatever the weather.

Thames Water’s action on leakage

We acknowledge that our average leakage level has been higher than our targets over the last few years. That’s why reducing leakage is a top priority for us and we’ve got a strong focus on our performance. This is contributing to our improving results, with our leakage level in February 2020 being 63 million litres per day (Ml/d) better than our upper forecast for the month. This means that leakage in February 2020 was 21% lower than a year ago. Read about our detailed February 2020 performance and see our 2019/20 monthly and annual average leakage levels at a glance.

Our latest performance is in the context of the 27% reduction in leakage that we’ve delivered over the past 15 years. We know you want us to do even better, so we have ambitious plans to do this. Our plans focus around two key areas. First, innovation and improving the productivity of our leak repair work. Second, improving our data to help us work smarter.

Innovation and productivity

We’ve increased the resources we’re putting into tackling leakage, which means we’re spending more than £1 million every day on leakage prevention and maintenance of our network. We increased the number of teams dedicated to fixing leaks by 21% in 2018/19, which delivered a 22% increase in fixed leaks. Since April 2019, we recruited even more repair teams and changed the way we work to improve the productivity of those teams, meaning that so far in 2019/20 we have fixed an average of almost 1,400 leaks every week.

Of the total number of leaks we fix, we know that around a quarter are leaks at our customers’ properties. While these are private leaks, which customers are responsible for fixing, we continue to work with our customers to repair their leaks, which in some cases we can do free of charge.

Fixing leaks is only part of the story. We can only fix leaks when we know where they are and as the majority of leaks we fix are hidden, our first challenge is finding them. To improve our efficiency and accuracy in doing this, we’re developing two digital data-driven tools. The first identifies our leakiest water supply zones and the second helps us locate the leaks in those zones more quickly. The tools automate the analysis of multiple dynamic data sources, using algorithms and models to support our leakage analysts and technicians. The output from these tools is a clear and simple visual display of leakage information, including an up-to-date map of our network, colour coded to show the highest-risk pipes and pipe repair history.

To help make our repair processes even more efficient, we’ve brought in new ways of working and have created a data-driven performance management tool to help supervisors get the most from their repair teams. This involves a more formal daily meeting structure between supervisors and scheduling teams, a standardised process and clarified responsibilities, as well as a bespoke ‘performance dashboard’ for supervisors that brings together key information about repair teams’ performance. Together with improvements to planning and scheduling of work, this enhanced performance management regime is expected to deliver a 12% efficiency improvement in leak repairs.

We’re also constantly looking for new technology and approaches to help us in our mission. For example, we’re doing this through:

  • Special in-pipe cameras to help locate difficult leaks.
  • ‘Fingerprinting’ our water zones to create a baseline and help us better understand where and why leakage occurs
  • Trialling new leakage detection techniques including Snipe the leakage sniffer dog

Improving our data

To help us make good decisions on resources and leakage reduction initiatives, it’s really important that we have the most accurate and up-to-date data possible. This is because our leakage figures are affected by more than just the water that leaks from our pipes.

A key part of updating our leakage data is refining the assumptions we make about water use by unmetered households, which is affected by factors such as population growth and development of housing stock. In the past we made these updates on an annual basis, but we’ve now changed the process to a six-monthly cycle and we’ll look to increase this frequency to quarterly from April 2020.

As part of this change, we’ve also taken a ‘big data’ approach to our analysis and have brought in new data to cross-reference against the information we already hold. This includes using the data from our smart meters to better understand the overall trends in our customers’ water use. We can then use that to inform our assumptions about water use by unmetered households. This is only possible because smart meters give us 24 readings each day, compared to the one or two readings per year that we would have received for the same customers before their smart meters were installed.

Updating our leakage data in this way is important as it will help us improve our leakage reduction capability and delivery. This is because having better data on where leakage is happening allows us to concentrate our detection and repair efforts on the parts of our network where we can reduce leakage the most.

Preliminary indications from this work for the first half of 2019/20 suggest that we’ve underestimated unmetered household demand, which means we’ve overstated our leakage figures. We’ve therefore retained the estimated 20 Ml/d provisional adjustment to our leakage figures (backdated to April 2019) that we introduced to this report in autumn 2019. As noted in previous reports, this work is still subject to further validation and external audit.

In addition to the benefits noted above, we’re now using data from over 398,000 smart meters that are consistently providing us with better information about water use at all types of properties. For example, they highlight properties with a continuous flow of water, which is an indication of a possible leak. Using data in ways like this has helped to prevent over 31 Ml/d of leakage since April 2019.

We’ve also installed 27,000 acoustic loggers to date, which has helped us detect over 60 Ml/d of leakage since April 2019.

Get involved

You too can play an important part in helping us achieve our targets.

Feedback: We’d like your feedback on our leakage performance information. Please get in touch to share your feedback, or to discuss how we may be able to make this report more accessible for you.

We’ve also started a conversation thread on our community platform to create further opportunities for you to talk to us and to each other, and share ideas on reducing leakage. We’re keen to engage with you in a meaningful way and we intend to use your feedback to improve our understanding of how leakage affects you, so that we can adapt our services to better meet your needs.

Leakage reduction - at a glance

Monthly leakage 2019/20

Ml/d Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
Monthly leakage level 621 612 593 607 591 605 586 615 630 - - -
Upper forecast monthly leakage 646 614 592 582 571 560 561 574 623 666 662 594
Variance -5 -25 -2 +1 +25 +20 +45 +25 +41 +7
Ml/d Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
Monthly leakage level 621 612 593 607 591 605 586 615 631 612 599 -
Upper forecast monthly leakage 646 614 592 582 571 560 561 574 623 666 662 594
Variance -25 -2 +1 +25 +20 +45 +25 +41 +8 -54 -63

Year-to-date average leakage 2019/20

Ml/d Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
YTD average leakage 621 617 610 609 606 606 602 604 607 - - -
Upper forecast YTD average leakage
630 618 609 601 594 590 588 592 599 605/td> 604
Variance -25 -13 -8 0 +5 +12 +12 +16 +15
Ml/d Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
YTD average leakage 621 617 610 609 606 606 602 604 607 608 607 -
Upper forecast YTD average leakage


630 618 609 601 594 590 588 592 599 605 604
Variance -25 -13 -8 0 +5 +12 +12 +16 +15 +9 +2


1. The figures in the above tables are operational leakage data and therefore based on information from Thames Water source systems at a point in time. This data may be subject to a data refresh on a monthly basis and for year-end reporting.

2. The actual leakage level figures reflect the provisional 20 Ml/d adjustment from April 2019 to date relating to underestimated unmetered household demand (referenced in this report).

Conclusion of Ofwat investigation – June 2018

Ofwat investigated our leakage performance in 2017. In August 2018, we agreed a package of financial and non-financial commitments with Ofwat under Section 19 of the Water Industry Act 1991. We refer to these as our Section 19 Undertakings.

As part of our Section 19 Undertakings we agreed to pay £120 million back to our customers. This money comes solely from Thames Water’s shareholders and will be reflected in customer bills. In order to clarify and improve how we report leakage we appointed Victoria Borwick, former Deputy Mayor of London and former MP, as an independent monitor of our monthly leakage reporting. Victoria brings highly relevant experience of reporting and communicating complex matters to the public. We are working with Victoria to provide external and independent feedback and challenge to how we engage with and improve customer understanding of our leakage targets and performance against these.

Tonia Lewis continues to fulfil the role of Undertakings Compliance Officer. This role is responsible for ensuring overall compliance with the Undertakings and providing regular updates to our Board and Ofwat. We provided our third 6-monthly review of our compliance against our Undertakings to Ofwat on 31 October 2019 and met with them on 2 March 2020 to discuss the report and our performance further.

Read more

Helpful links

Find out more

Please see the further sections below for some helpful background information on leakage, as well as more detail about our performance and our leakage reduction plan.

For all water companies, leakage is made up of any water that they are unable to account for as having been used by someone. As you would expect, this includes any water that is lost from the pipe network into the ground – ‘true’ leakage. However, it also includes any water use of which a water company isn’t aware, such as illegal use and any higher-than-estimated use by households and businesses that don’t have a meter.

We supply almost one third of the water used by people and businesses in England and Wales each day, to ten million customers, through 31,550 km of water pipes.

To estimate leakage, like all water companies we compare the measured volume of water we put into supply against the volume we estimate is being used. The difference between these two values is what we record as leakage. We measure leakage in millions of litres per day (Ml/d). As an idea of scale, an Olympic-size swimming pool (50m x 25m x 2m) contains 2.5 million litres of water.

There are many factors that affect ‘true’ leakage, but the three main ones are:

  • Natural wear and tear on our network – if we were to carry out no work, other than repairs of visible leaks reported to us by our customers, we estimate that leakage would increase by 322 Ml/d over the year due to the wear and tear buried pipes experience from ground movement and pipe corrosion, as well as the repeated stresses to which they are subjected.
  • The seasons – in typical weather conditions, the colder winter months increase leakage due to pipes shrinking, causing the joints between them to open up. Equally, warmer temperatures can undo this effect to reduce leakage. This is expected to even out over the year but leads to peaks and troughs in leakage levels.
  • Extreme weather events – can impact leakage over and above the typical seasonal rise and fall. A freeze followed by a rapid thaw can over-stress pipes causing them to burst. Similarly, prolonged hot dry periods can shrink the earth that supports pipes, causing them to move and break.

We’ve set annual leakage targets up to 2020. Leakage levels change throughout the year, therefore to measure and report our performance we take our daily leakage figures (in Ml/d) and average them for each month and in turn across the year.

For the last 4 years actual leakage compared with our targets is shown in the table below. We beat our target in 2015/16 but have missed it for the last 3 years. The results have been externally audited and confirmed.


Our target for 2019/20 of 606 Ml/d requires a reduction of 12% from our 2018/19 annual average. This reduction would be the highest level of reduction seen since 1999/2000. We’re committed to doing all we can to hit it.

We continue to do all we can and, as we wait for the operational leakage figures for the final month of the reporting year to become available, our confidence in achieving 606 Ml/d is growing. Our central forecast is that we will achieve our 2019/20 target. However, this remains subject to the further validation and external audit work that is underway on the update of our leakage data in relation to unmetered household demand.

Leakage reduction remains one of our top priorities and we continue to strive to deliver our 606 Ml/d target. Keeping our stakeholders updated about our leakage performance and forecasts also remains a priority.

We also remain committed to reducing leakage by over 15% from our 2019/20 target level by the end of 2025 and by 50% in the longer term.

Headlines for February 2020

  • Our leakage level for February was 599 Ml/d, which is 21% lower than February last year.
  • We completed 5,532 leak repairs in the month.
  • We repaired leaks which we estimate have prevented leakage equivalent to 31 Ml/d of water in February.

Leakage level

The graph below shows the actual amount of water lost from our network in February 2020, compared with the upper and lower forecasts in our leakage reduction plan. In February our leakage level was 599 Ml/d, which is 10% lower than the upper forecast level in our leakage reduction plan.

This improved performance has been driven by several factors. First, over the year to date we have increased the number of repair teams we have to an average of nearly 200. This has meant that we have been able to bring down our backlog of repairs to mains pipes to around 120 jobs – the lowest level in over a decade. In turn, this has helped us to reduce the time we take to repair non-visible leaks to the best performance in the last 8 years.

In addition to this, we have been helped by the milder weather conditions experienced so far this winter. Leakage increases as water temperature drops because the pipes contract, opening up joints and increasing the size of existing leaks. The milder-than-average conditions mean that this hasn’t happened to the extent that we forecast in our plans. Another effect of the milder weather is that we haven’t seen the increase in visible leaks that we would normally expect at this time of year. This in turn has allowed our repair teams to focus on existing non-visible leaks, which provide the biggest leakage reduction benefit.

Note: The actual leakage level figure reflects the provisional 20 Ml/d adjustment relating to underestimated unmetered household demand (referenced in this report).

Leaks fixed

We fix three different types of leak. On average:

  • About half are non-visible leaks - these are below the surface and aren’t easy to find. We find them using data, technology, and people. Fixing non-visible leaks has the biggest impact on reducing leakage.
  • About a quarter are visible leaks - these are easy to see, for example on pavements or roads.
  • About a quarter are customer leaks - these are within the boundary of a customer’s property up to the point where the pipe enters their home/building.

As shown in the graph below, we fixed 5,532 leaks in February 2020. This is slightly below our overall plan for the month. However, we fixed 13% more non-visible and customer leaks than our plan, so the shortfall of repairs was driven only by the number of visible leaks. This was as a result of the milder weather experienced through February,which as explained above meant we had fewer visible leaks to fix – for February there were 49% less than our plan.

Since April 2019 as a whole, there have been fewer visible leaks than we forecast which has allowed us to focus our repair efforts on non-visible and customer leaks. Overall, including all types of leaks, we’ve fixed 66,477 leaks since April 2019 which is a 2% improvement on the same period last year. This has been achieved by recruiting more people to fix leaks as well as making improvements to our planning processes.

Leakage reduction

The graph below shows the estimated volume of water saved from fixing non-visible and customer-side leaks in February 2020, compared with our plan. We don’t include visible leaks because they are not considered to reduce the underlying level of leakage. We repaired leaks which we estimate have prevented leakage equivalent to 31 Ml/d of water in February. This was 1 Ml/d below our plan.

Despite repair numbers for non-visible and customer leaks being above target levels for February, the estimated volume of water saved was slightly below our plan due to the higher proportion of smaller leaks that have been detected and repaired compared to our plan.

Performance against our plan

You can see our 2019/20 monthly and annual average leakage levels at a glance here.

We believe there are several reasons for our year-to-date leakage level being slightly higher than our leakage reduction plan.

The largest component of leakage reduction is achieved though the repair of non-visible and customer leaks. We’ve repaired 7% more of these leaks than our plan during the eleven months from April 2019 to February 2020. This represents an 11% increase in the number of non-visible and customer leaks fixed over these eleven months compared to the same period in 2018/19. Despite this, the leakage savings delivered were slightly behind plan because the mix of leaks is different to our forecast which is based on historical performance. As the mix of leaks fixed contains a lower proportion of large leaks and more smaller leaks, this provides a smaller average leakage reduction benefit per repair.

The leakage reduction values above include an adjustment for the estimated volume of water saved from our non-visible and customer leak repairs. We applied a downward adjustment of 16% from April 2019 while we continued to analyse and understand the results of leakage repair work to date. In mid-August we began to analyse our repair performance data on a more frequent weekly basis to give us a more accurate view of the estimated volume of water saved. Due to improvements in our processes, the accuracy of how our leakage repair jobs are classified has improved significantly with the weekly adjustment getting close to zero during February. On the basis of this enhanced analysis the adjustment applied has been an average of 4% from mid-August onward, although we retained the original 16% adjustment for the period from April to mid-August 2019.

Improving the accuracy of categorisation of work we carry out and how this is captured in the field is an area we’ve been working hard to improve. We believe that longer term there will be benefits for leakage targeting and our overall leakage reduction strategy to be gained from improving data quality that will then allow the application of advanced data analytics. We have a number of activities in place to address data improvements, such as issuing a guidance booklet to our repair teams that details the process for capturing repair activity.

We remain committed to doing all we can to address leakage and to achieve our 2019/20 target. This is a key area of focus for our Executive and Board. Our Leakage Task Force reviews our performance and reduction plans every two weeks to ensure we retain control over the way in which our leakage reduction plan is carried out.

We’ve taken account of actual performance to date and refreshed the remaining month’s work plan to help get us back on track. A suite of initiatives has been introduced to improve leakage detection and productivity of our repair work. We’ve increased our repair and maintenance resources, which means that for January and February 2020 we had an average of around 2% more repair teams than we had at that time last year. In the meantime, we maintain a dialogue about our leakage reduction plan with our key stakeholders.

In addition, we have a programme of initiatives aimed at improving the accuracy of measured and unmeasured customer water use. This includes recording usage patterns, meter accuracy and improving property/occupancy data.

Innovation in finding leaks

Fixing leaks is one of the key activities we undertake to reduce leakage. Because fixing non-visible leaks has the biggest impact on reducing leakage, we prioritise them whenever we can. But, to fix more non-visible leaks, we first need to find them on our pipe network. As we work to reduce leakage below our all-time lowest level, we have to be increasingly innovative in how we do this.

In 2018/19 we did this by:

  • Smart water meters - When we install a customer meter, we can immediately identify if there’s a leak. Our programme to install smart water meters is the second largest in the world and will help protect future water supplies.
  • Installing acoustic loggers - Acoustic loggers listen for the noise water makes as it leaks from pipes. They help us to improve the efficiency of leak detection. We installed approximately 27,000 loggers and are repairing the leaks located as a result of the information that the loggers are generating.
  • More leak detection technicians - We increased the number of leak detection technicians working for us, and we introduced a programme of field audits to validate their work. They use data to judge where leaks are occurring and tell repair teams where to dig.

This year we’re increasing the number of leaks we find through:

  • ‘Fingerprinting’ our water zones - We’re using key data about each of our water zones to begin to create unique fingerprints for each of them. That in turn will help us to understand the drivers of leakage performance and burst mains through various analysis methods.
  • Temperature analysis – Working with a number of other water companies, we’re in the early stages of developing an innovative leak detection method that aims to use temperature data to pinpoint areas with larger leaks that are difficult to detect.
  • Leak analysis – We’re starting to look at how soil conditions can cause visible leaks, so that we can build up a better understanding of the environmental conditions that lead to leaks.
  • Digital detection tools – We’ve been developing two core products to help us detect leakage more quickly and efficiently by first targeting the right areas of our network, and then focussing on the right streets within those areas.

Leakage reduction activities

To help us reach our target by 2019/20, in addition to leak repairs we have additional leakage reduction activities taking place. These activities are intended to generate a sustainable reduction in our leakage levels and they include:

  • Pressure management - We manage the water pressure within parts of our network to reduce the likelihood of pipes bursting. This also reduces the amount of water lost through existing leaks.
  • Mains replacement - We have a long-term programme of replacing or repairing worn pipes, and we prioritise the ones which may cause most disruption if they burst or leak.
  • Accounting for all water used - We're continually trying to improve our estimate of water used from our network. This activity focuses on stopping illegal use (for example unlicensed standpipes and illegal connections), replacing faulty customer meters, and ensuring all properties are registered on our billing system correctly.

Risks to our plan

The biggest risk to us meeting our target in 2019/20 is the weather. Both extreme cold and hot conditions can increase leakage levels. To mitigate this risk, we continue to review and update our leakage reduction plans against a range of short and long-term weather scenarios. We formally review our plans every three months to assess performance and identify areas where we need to undertake additional activity. This work enables us to review the resource requirements and ensure they are sufficient to detect and repair the number of leaks required.

The size and scale of the plan in place to achieve our highest level of leakage reduction since privatisation makes it more difficult to fully understand and estimate the water savings made from fixing leaks and the impact of the different work mix. Our level of understanding improves with each month as we build experience and adapt our plans.

The water saved from repairs appears to be lower than we’ve experienced historically. As leakage levels reduce to new all-time low levels it is likely that the average leak is becoming smaller. Assuming this is the case, the number of leaks we need to repair to achieve a given reduction in leakage has to increase. To address this risk, we increased repair activity targets in our quarterly reforecast, supported by the recruitment of 9% more repair teams between August and September 2019 to help deliver this.

Other material risks that we monitor are the effectiveness of our leak measurement models, our understanding of how our pipes react to rapid changes in weather, and our ability to pre-empt network deterioration. Our Leakage Task Force is addressing these and other questions to ensure our planning assumptions and leakage reduction strategy are effective. The work of the Task Force will further help us refine our plans for 2019/20 and beyond.

As mentioned earlier in this report, an important aspect of enhancing the accuracy of our calculation of leakage is refining the assumptions we make regarding water use. Currently the analysis carried out has identified that we have over-estimated our leakage by 20 Ml/d and, as a result, a provisional adjustment was applied to our leakage figures in autumn 2019 (and backdated to April 2019). We acknowledge this is still subject to further analysis and validation as well as approval from our external auditors.

To ensure effective control of the above risks we’ve allocated senior management resource to drive performance in key areas we know are challenging, such as, leakage detection and repair output performance, data capture, analysis of leaks fixed and job auditing.

The information provided is, to the best of our knowledge, accurate, but may be subject to change from time to time. The information does not represent formal annual regulatory reporting.