MC

Ofwat · open consultation · 27.08-09.10.2026

Greater uncertainty can shrink reported leakage

Water leakage is not measured directly. When the balance does not close, MLE allocates the gap across its components. Ofwat says greater uncertainty on distribution input can leave less of that gap assigned to leakage, affecting payments and reputation without a physical pipe being repaired.

Primary source Ofwat, 27 August 2026, pp. 16 and 41-42

England and Wales · incumbent wholesale water companies · PR24 2025-30 · NAVs excluded

This is a feature of the method identified by Ofwat, not a finding about any company's conduct.

The same physical network

Two allocations of the gap

Conditional, not-to-scale schematic. The pipe stays fixed; the statistical allocation described by Ofwat changes. Direction and magnitude depend on the full balance. This is not a company water balance.

01 / the measurement problem

Leakage is not measured directly

Ofwat treats total leakage as a balancing item estimated through two paths, rather than a directly observed sector total. The estimates are compared; an inevitable, ideally small water-balance gap remains.

Top down

Distribution input minus consumption and the other accounted uses.

Bottom up

Night flows in metered areas, completed with the other leakage components.

input = consumption + uses + leakage + gap
When the estimates do not match, a water-balance gap remains. It cannot simply be ignored.

02 / reconciliation

The gap has to be allocated

MLE, or Maximum Likelihood Estimation, adjusts the balance lines until input and outputs match. Ofwat describes two factors in that allocation: the size of each component and the uncertainty with which it is measured.

  1. A
    Calculate the gap. Measured input is compared with the sum of consumption, uses and estimated leakage.
  2. B
    Give each line an uncertainty band. Larger or more uncertain components can receive a larger share of the gap.
  3. C
    Publish the post-MLE annual average in Ml/d. The PR24 commitment then measures the percentage reduction in the three-year average from the 2019-20 baseline.

03 / the effect

The pipe stays fixed. The ledger moves.

In many cases, Ofwat says, greater uncertainty on distribution-input meters assigns more of the gap to input and less to leakage.

This is the structural effect identified by the regulator. It does not show that any company deliberately increased uncertainty.

Above ±2%The balance gap is non-compliant and the company must explain it to Ofwat.

Beyond +5% or -5%The gap is too wide for a valid MLE adjustment and triggers a review of material components.

Independent assuranceConfidence intervals and the balance require third-party assurance; ODI payments must reflect real performance, not method or data changes.

04 / order of magnitude

The intervals are not small details

Distribution input is the largest line in the balance. Ofwat says reported leakage averages 19% of distribution input across the incumbent wholesale companies in England and Wales covered here. A few percentage points of input uncertainty can matter for a component of that size.

Distribution input
current guidance2%-4%
reported by companies±2% to slightly worse than ±5%
Detailed, reliable estimates
current definition, including distribution leakage8%-12%
observed cross-company range for bottom-up leakage8%-25%

Wider intervals can describe honest measurement limits. Ofwat's concern is that methods differ across companies and that the interval affects the reconciled value. The consultation does not expose the reporting year, weighting method, or company values behind the 19% aggregate or the 8%-25% range, so neither can be reproduced from that document alone.

05 / claim boundaries

What the document does not prove

It does not prove fraud. The consultation identifies an incentive in the method, not a company's intent.

It does not make leakage directly observable. Even a better method remains an estimate built from measurements, samples and assumptions.

It does not say MLE should be removed. Ofwat's preferred option is to retain reconciliation with clearer, tighter confidence-interval rules.

It does not publish a company audit. Without complete pre- and post-MLE values, this project attributes the effect to no operator and estimates no financial amount.

06 / rule status

The future method is still under consultation

Ofwat opened the consultation on 27 August and accepts responses until 9 October 2026. The document does not change targets or payment rates in the current regulatory period.

Preferred optionRetain MLE with clearer, tighter requirements for calculating confidence intervals.

Connected changeA common standard for selecting, installing, maintaining and checking meters.

HorizonTighter MLE guidance could be implemented within one year. Metering best practice could take up to two years to develop, with company adaptation in later years.

Read Ofwat's consultation

07 / evidence trail

The sources behind the mechanism

  1. Ofwat, Leakage Reporting Methodology - Consultation August 2026. See pages 2, 4, 9-10, 16, 41-42, 45-46 and 50 for scope, method, ranges, timelines and population.
  2. Ofwat, PR24 Leakage Performance Commitment definition, version 2.0 Final-determination definition dated 19 December 2024 and republished 31 March 2025; see pages 3-4, 23 and 25-26.
  3. Ofwat, PR24 final-determination performance commitment definitions Controlling index for the 2025-30 period and the current leakage definition.

Last checked 28 August 2026. Claims and ranges come from the official documents. The diagrams are original explanatory work.