Data Quality and Indicator Sensitivity in Water-Loss Benchmarking: An Exploratory Study of a Purposive Sample of Eleven Greek Water Service Providers

Abstract

Water-loss indicators answer different management questions, yet they are often compared as interchangeable rankings. This exploratory study asks what regulatory water-loss reporting can support when the inputs behind each indicator are unevenly documented. For a purposive sample of eleven Greek providers reporting under a harmonised framework, the 2024 and 2025 IWA water balances were reproduced, volumetric indicators were normalised to the hours under pressure, and the twelve Infrastructure Leakage Index (ILI) values reported by seven providers, or reproducible from their WB-EasyCalc files, were examined. Sensitivity to metering, pressure, and denominator inputs was quantified, and each ILI case was screened for data adequacy. In 2025, non-revenue water ranged from 26.7% to 71.1% of system input volume, with a mean of 46.2%. Six providers reported an identical percentage in the reference table for both years; in four, the leakage level in that table differed from the real losses in the same provider’s balance by 3.8–17.5 percentage points. Average pressure was stated for all seven ILI cases but measured in none, with three adopting the same round value. The study documents failure modes in indicator reporting rather than estimating their prevalence, and proposes a tiered indicator set published with the documentation that makes each indicator readable.

1. Introduction

Non-revenue water (NRW) remains a major operational and financial concern for water utilities. Global estimates place the annual volume at more than 100 billion m³, while recent European evidence shows large variation among countries and utilities [1,2]. A standard water balance is therefore essential, but the indicator selected from that balance determines what is compared. A percentage of system input volume is readily understood; it is also influenced by consumption, exports and the completeness of billed consumption. A technically focused indicator may reduce those denominator effects while introducing requirements of its own.

The International Water Association (IWA) water balance separates authorised consumption from apparent and real losses and recommends a family of indicators rather than a single score [3,4,5,6]. The Infrastructure Leakage Index (ILI) is the ratio of Current Annual Real Losses (CARL) to Unavoidable Annual Real Losses (UARL). UARL is calculated from mains length, number of service connections, private-pipe length, and average pressure [7,8]. In principle, ILI compares current leakage-management performance with a system-specific technical reference. NRW%, by contrast, relates unbilled water to system input and is useful for resource and revenue oversight [9]. The two indicators should not be expected to answer identical questions.

Their interpretation also depends on source data. Top-down water balances are sensitive to bulk-metre error, customer-metre under-registration, and estimated consumption [10,11,12]. ILI additionally depends on the completeness of network length, the definition of service connections, metre location, and a defensible average pressure. Application guidance has evolved since the original UARL publication. By 2009, the earlier lower limit of 20 connections/km had been removed, and the minimum-size check was expressed as Nc + 20Lm > 3000; connection density remained relevant primarily to the choice of a per-connection or per-kilometre volumetric indicator [13]. More recent guidance recommends a system correction factor (SCF) when size, pressure, burst behaviour or pipe materials make the standard UARL a poor approximation [14,15]. Therefore, it is inaccurate to treat a fixed density threshold as an automatic test of ILI validity.

Pressure is an intended part of UARL, not a nuisance variable. Leakage varies with pressure and pipe response, and the FAVAD literature shows that the pressure exponent can depart substantially from one [16,17,18,19]. This makes measured, zone-weighted pressure preferable to a single engineering estimate. Intermittent supply adds another layer: CARL, customer-metre error and time-normalised indicators become sensitive to filling, emptying and the time for which the network is pressurised [20,21,22,23,24]. In such systems, indicator values require an explicit supply-regime label.

The regulatory setting is changing. Directive (EU) 2020/2184 requires leakage assessment using ILI or another appropriate method for providers above the Directive’s size thresholds [25]. Italy’s technical-quality framework reports both percentage and linear losses, with linear losses expressed in m³/km/day, while Ofwat publishes leakage performance in absolute and normalised volumetric forms [26,27]. These examples favour transparent indicator sets over a universal percentage threshold.

Greece now provides a useful, though not nationally representative, setting for examining these issues. Law 5037/2023 extended the remit of the Regulatory Authority for Waste, Energy and Water (RAAEY) to water services, and the managerial-adequacy process introduced common documentation templates [28,29,30,31]. Earlier Greek studies have documented both urban-network performance and the constraints facing small settlements and islands [32,33,34]. The present study uses the dossiers of eleven providers as a purposive multi-provider benchmark. It asks (i) what the 2024 and 2025 water balances show descriptively; (ii) how strongly metering practice can affect NRW%; (iii) how the reported ILI values should be interpreted when the denominator inputs are documented to different degrees; and (iv) what reporting scheme is defensible with the information already collected. The study is explicitly exploratory. It does not claim a national test of ranking agreement.

The contribution of this study is evidential and procedural rather than methodological. It introduces no new indicator and no modification of the UARL equation. Four elements distinguish it. First, it examines the traceability of indicator values within linked provider dossiers, following each value back to the field and document from which it was read. Second, it compares fields directly where the reference year and the reporting boundary coincide, and states explicitly where that correspondence cannot be established—as it cannot for the ILI case set, which combines two reference years from the regulatory reference table with two cases reproduced from WB-EasyCalc files, one of them for an earlier year. Differences in source and year are not excluded in advance as an explanation of divergence; they are part of the problem under study. Third, the documentation of the denominator—mains length, connection count, private-pipe length, average pressure and hours under pressure—is treated as an object of analysis rather than as a set of given inputs, and the movement of each indicator under perturbation of those inputs is quantified. Fourth, the documentation gaps observed are converted into explicit reporting requirements rather than into performance judgements.

2. Materials and Methods

Figure 1 summarises the research process, the two analytical samples it produces and the steps through which the reporting scheme of Section 3.6 was derived.

2.1. Study Design, Data Source, and Sample

The source material comprises documentation dossiers prepared for the Greek managerial-adequacy framework [28,29,30,31]. Eleven retained providers had complete water-balance tables for both the 2024 and 2025 cycles and form the benchmark sample. Two data streams were used for leakage. The first is the reference Table B2 of each dossier [28,29,30,31], which records the reported non-revenue water of the internal network (field D.1.4), the reported leakage level (field D.2.1) and, for some providers, the ILI with its CARL and UARL (field D.2.2), separately for 2024 and 2025. Field D.2.2 carries the resulting figures but not the inputs behind them: it refers to a separate leakage-assessment report held in sub-folder 06 of the same dossier, and that report holds the main line length, connection count, average pressure, and private-pipe length from which the index was computed. Five providers—Aktio-Vonitsa, Alonnisos, Fournoi Korseon, Leros and Sifnos—reported an ILI for both reference years in that table. The second stream is the WB-EasyCalc file (version 6.17; Liemberger & Partners, Pressbaum, Austria), available with sufficient fields to reproduce an ILI case for Chios (2023) and Souli (2025) [35]. Seven providers and twelve provider-year values therefore enter the ILI analysis, while the four providers that contribute no ILI case—Dorida, Korinthos, Limnos and Trifylia—report leakage only as a percentage or as an annual volume per kilometre of mains. No provider outside this retained set enters either the water-balance benchmark or the ILI analysis. A value from one year was never combined with infrastructure data from another year. A third component of the same dossier, the Annex II reference table (Table B3) [28,29,30,31], reports the transmission and the distribution network length, the number of connections, and the population served for the 2025 reference year. It is used here for provider context and for cross-checking derived quantities, and supplies the context columns of

Table 1. Analytical samples, provider context, and reference years.

Provider	Provider Type	Setting	Supply Regime	Population Served (2025)	SIV 2025 (m3/Year)	Service Connections (2025)	Mains Length (km)	ILI Case (Year, Source)
Aktio-Vonitsa	Municipality	Mainland, coastal	Continuous	14,644	5,096,115	10,000 a	nd	2024, 2025 (B2)
Alonnisos	Municipality	Island	Continuous	3105	361,018	2000 a	26.3	2024, 2025 (B2)
Chios	DEYA	Island	Intermittent (4 h/day; 2023 case)	51,936	4,229,898	35,861	1000.0	2023 (WB-EasyCalc)
Dorida	Municipality	Mainland, mountainous	Not stated	12,034	1,207,440	9500	200.0	—
Fournoi Korseon	Municipality	Island	Continuous	1340	201,610	1000 a	5.4	2024, 2025 (B2)
Korinthos	DEYA	Mainland, coastal	Not stated	55,941	5,906,214	23,559	534.3	—
Leros	Municipality	Island	Continuous	7907	1,419,650	4313	60.0	2024, 2025 (B2)
Limnos	Municipality	Island	Not stated	16,411	1,827,900	13,900	550.0	—
Sifnos	Municipality	Island	Continuous	6600	553,500	3675	115.0	2024, 2025 (B2)
Souli	Municipality	Mainland, semi-mountainous	Continuous	11,200	1,287,740	7019	280.0	2025 (WB-EasyCalc)
Trifylia	DEYA	Mainland, semi-mountainous	Not stated	22,431	3,222,224	20,600	299.0	—

Note: All eleven providers contribute to the 2024–2025 water-balance benchmark. DEYA = Municipal Water Supply and Sewerage Enterprise. B2 = ILI reported in the Table B2 regulatory reference table; WB-EasyCalc = ILI reproduced from the provider’s WB-EasyCalc file; an em dash indicates that the provider contributes no ILI case. Population served, service connections, and mains length are the 2025 values of Annex II (Table B3) of each dossier and are reproduced as submitted; the population served is contextual and is not used as an analytical denominator anywhere in this study, and it is not the administrative population of the municipality. System input volume is the 2025 value reported in the provider water balance. The mains length shown here is the distribution network and excludes transmission mains, which Annex II reports separately; the mains length adopted in each ILI case follows the convention of its own source and does not always coincide with this column, as Section 3.5 sets out. For Chios, the supply regime and the infrastructure inputs used in the ILI case refer to the 2023 WB-EasyCalc file, whereas the columns of this table refer to 2025. nd = not stated in the material used for this study. ^a Reported in the dossier as an estimate or as a round figure.

; its values are reproduced as submitted and are not used as analytical denominators. The leakage-assessment reports and the WB-EasyCalc files that take their place for Chios and Souli were read alongside the reference table. Appendix A records, for every denominator input used here, the document it was read from and whether the source presents it as a measurement or as an assumption.

Sampling was purposive and based on dossier availability. The common templates improve definitional consistency, but they do not make the sample representative of all Greek providers and do not guarantee that submitted values were independently measured; this distinction is central to responsible benchmarking [36]. Provider names are retained as requested by the data holders. Provider context in Table 1 is taken from Annex II of the same dossiers rather than from administrative statistics, because the population served by a network and the administrative population of a municipality need not coincide. The 2021 census [37] was used only to check that the reported served populations are of the expected order, and population is not used as an analytical denominator anywhere in this study. Table 1 identifies which providers contribute to the balance benchmark and which contribute an ILI case, and with what reference year and source; Figure 2 shows their location.

2.2. Water-Balance and Volumetric Indicators

The submitted balances follow the IWA structure [3,4]. System input volume (SIV) is divided into authorised consumption and water losses; losses are divided into apparent and real losses. NRW was calculated from billed authorised consumption (BAC):

NRW (%) = (SIV − BAC)/SIV × 100

(1)

Real losses were also expressed per service connection and per kilometre of mains. For intermittent service, both indicators were normalised to the period for which the system was pressurised:

qRL = CARL × 1000/[Nc × (T/24)]   [L/connection/day, w.s.p.]

(2)

qL = CARL/[Lm × (T/24)]   [m³/km/day, w.s.p.]

(3)

Here, CARL is the mean daily real-loss volume (m³/day), Nc is the number of service connections, Lm is mains length (km), T is the mean number of hours per day for which the system is pressurised, and w.s.p. means “when system pressurised”. Equation (3) uses m³/km/day, consistent with the unit used by ARERA [26]. For continuous systems, T = 24. Metering coverage of billing was calculated as billed metered consumption divided by the sum of billed metered and billed unmetered consumption. Where the dossier reported real losses only as an annual volume per kilometre of mains, mains length was recovered by dividing the annual real-loss volume by that value, annual volumes being obtained from mean daily values on a 365-day basis throughout; the recovered length is a derived quantity rather than a measurement and is identified as such wherever it is used.

2.3. ILI Calculation and Sensitivity Analysis

The standard UARL and ILI equations were reproduced as follows [7,8,13,14,15,38]:

UARL (L/day) = (18Lm + 0.8Nc + 25Lp)P

(4)

ILI = CARL/UARL

(5)

Lp is the total private-pipe length from the property boundary to the customer metre (km), and P is the average operating pressure (m). Private-pipe length is stated in every ILI case once the leakage report or the WB-EasyCalc file is read alongside the reference table, and the values adopted differ by an order of magnitude. Leros, Alonnisos and Fournoi Korseon take it as negligible, on the stated ground that metres sit at the property boundary, as do the Sifnos leakage reports; the basis behind the ILI that the Sifnos reference table reports uses 15 m per connection instead (Section 3.5). Chios and Souli use the 15 m per connection that WB-EasyCalc offers as a default, giving 534.8 and 105.3 km of private pipe, respectively; for Chios, that total corresponds to the best-estimate connection count of 35,656 that the file also carries rather than to the 35,456 registered connections used for the volumetric indicators, and the reported UARL follows the former. Aktio-Vonitsa states 40 km, or 4 m per connection, described in its report as typical for settlements of that density. Recomputing UARL from the inputs adopted in each case (Section 3.3) reproduces the reported denominator to within about 1% in every case; the residual arises from rounding in the displayed inputs and, for two providers, from inputs that differ slightly between the two reference years. Appendix A sets the reported and the recomputed values side by side. The exercise confirms the arithmetic and adds no independent evidence: the private-pipe term is an assumption in all seven cases, and depending on which assumption is used, it contributes between nothing and about 17% of UARL. The consequence is clearest in Aktio-Vonitsa. The UARL implied by its reported ILI, 440 m³/day, is what the standard equation returns for 200 km of mains, 10,000 connections, 40 km of private pipe, and 35 m of pressure, the midpoint of the 30–40 m range its report states; the same equation gives 406 m³/day with no private pipe and 537 m³/day at 15 m per connection. Adopting the latter would lower the 2025 index from 21.4 to 17.5. The dossier separately states a scenario range whose lower bound is 17.4; that bound and the 17.5 reproduced here from the displayed inputs are close but not identical and should not be treated as the same quantity. For Chios, the reported UARL of 503 m³/day is the 24 h value of 3019 m³/day scaled by the four hours for which the system is pressurised. CARL and UARL are therefore both expressed per calendar day, and their ratio is internally consistent; an index formed by dividing the same CARL by the unscaled 24 h value would compare a four-hour system against a continuously pressurised reference, and is not reported here. Equations (4) and (5) were evaluated after CARL and UARL were expressed in the same units; UARL was obtained in L/day from Equation (4) and converted to m³/day before the ratio in Equation (5) was formed. Where UARL was instead recovered as CARL divided by the reported ILI, the result was the denominator implied by the provider’s own submission and was a restatement of that submission rather than an independent verification of the reported index.

Sensitivity was examined descriptively. For Souli, ILI was recalculated over pressure scenarios of 30–70 m while holding CARL and other inputs fixed. The dossier value of 60 m is treated as an unverified assumption, not as a measured mean pressure. A one-at-a-time ±20% perturbation was also applied to P, Lm, Nc, and private-pipe length; each input was varied while every other input was held at its base value, so that total private-pipe length was held constant when the connection count was perturbed. These calculations measure numerical dependence on inputs; they do not demonstrate that the ILI is physically incorrect. The connection-density plots likewise illustrate the normalisation built into UARL. Density below 20 connections/km is used only to motivate reporting m³/km/day alongside L/connection/day, not to invalidate ILI [13]. The 2009 minimum-size expression Nc + 20Lm > 3000 was evaluated for every ILI case and is reported with the inputs. All water balances, indicators and sensitivity calculations were reproduced, and all figures were prepared, in Python (version 3.12; Python Software Foundation, Wilmington, DE, USA) with the NumPy (version 2.1), pandas (version 2.2) and Matplotlib (version 3.9) libraries.

2.4. Data-Quality Assessment and Analytical Scope

No inferential test of ranking agreement was performed. The eleven providers supply a descriptive benchmark rather than a random sample, and the ILI cases differ in the completeness of their denominator documentation; one of them (Chios) is from an earlier reference year and operates under intermittent supply. A rank-correlation coefficient computed on this material would be driven by documentation quality as much as by physical performance. Similarly, Bland–Altman analysis of normalised ranks would not evaluate agreement between measurements on a common scale. Results are therefore reported as ranges, comparisons between the two reported cycles, case contrasts and sensitivity scenarios. Any comparison between reference years concerns the values entered in the dossier fields and is not a trend in measured leakage; no comparison of that kind is attempted across the ILI case set, whose records differ in source and in reference year.

Five data-adequacy domains were screened for each ILI case: network-length completeness, pressure documentation, supply-regime comparability, CARL/calculation consistency, and documentation of private-pipe length. The screen records issues evident in the submitted material; it is not an external audit. Aktio-Vonitsa was flagged because its 200 km mains length and its 10,000 connections are both stated as estimates, and no separate connection register is kept. Leros, Alonnisos, and Fournoi Korseon were flagged because the average pressure their reports adopt is the same round 40 m in all three and is described in one of them as an assumption of the analysis, and because the private-pipe length is taken as zero. Fournoi Korseon was flagged additionally because CARL, UARL, and ILI are identical in both reference years. Sifnos was flagged because the reference table and the leakage reports of the same dossier compute the index from different mains lengths, private-pipe lengths, and pressures. Chios was flagged for intermittent supply. Souli was flagged because 60 m was not independently confirmed and because the CARL in the WB-EasyCalc file is not reconciled with the leakage level reported in the same dossier. Top-down CARL estimates and uniform component allocations were treated as general limitations [10,11,12,20,39,40].

2.5. Analytical Logic: Documented Failure Modes Rather than Prevalence

A purposive sample supports some claims and not others, and the distinction governs every result reported below. These eleven dossiers cannot establish how frequently any reporting problem occurs in the Greek water sector, because the providers entered the study through documentation availability rather than through random selection. They can establish that a problem occurs within the material examined and that it is material to the indicator value: a single submission in which a reported leakage level and the balance in the same dossier differ by 17.5 percentage points is sufficient to show that the field from which an indicator is read must be specified, and no larger sample is required for that conclusion. The study therefore documents failure modes in indicator reporting and measures their numerical consequences; it does not measure how often they occur, and it does not establish that the gaps documented here are the only ones such material can contain. Counts given in the text—six of eleven, four of seven—describe this sample and are not estimates of a population parameter.

A related distinction governs the reporting scheme proposed in Section 3.6. Its generality is not claimed to follow from the sample, and it is proposed rather than validated. Its conceptual basis is, first, the algebraic structure of the indicators themselves: NRW as a percentage, real losses per connection, real losses per kilometre and ILI differ in their denominators, and that difference is a property of the definitions rather than of any particular network. This structure supports the input requirements attached to each tier; it does not by itself demonstrate that the scheme is workable in practice. The second element of the basis is independent precedent for each requirement, since publishing losses in both percentage and absolute form, conditioning a loss indicator on metering coverage, reporting a normalised volumetric loss and grading the validity of audit inputs are all established practice in existing frameworks [5,25,26,27]. The contribution of the sample is narrower and empirical: it shows which of these safeguards address failure modes that are present in submitted regulatory material, and which inputs are the ones that are in practice missing. Section 4.5 states the conditions under which the scheme would be expected to transfer and the testing that remains to be done.

2.6. Use of Generative AI

During the preparation and revision of this manuscript, the authors used Anthropic Claude (Opus 5) to support English-language editing, stylistic refinement, structural organisation, consistency checks across the text, tables, and figures, and the preparation of selected visualisations and tables. This tool was not used to generate primary data, determine the study results, or make final methodological or interpretive decisions. All AI-assisted output was critically reviewed, verified against the underlying data and cited sources, and revised by the authors, who take full responsibility for the accuracy, integrity, and final content of the manuscript.

3. Results

3.1. Descriptive 2025 Benchmark

Across the eleven 2025 water balances, NRW ranged from 26.7% for Fournoi Korseon to 71.1% for Aktio-Vonitsa, with an unweighted mean of 46.2% (

Table 2. Water-balance indicators for the 2025 benchmark sample, ordered by NRW.

Provider	SIV (m3/year)	NRW (%)	Real Losses (%)	Apparent Losses (%)	Authorised Consumption (%)
Aktio-Vonitsa	5,096,115	71.1	67.5	2.7	29.9
Leros	1,419,650	66.5	60.5	4.5	35.0
Alonnisos	361,018	63.3	54.7	7.6	37.7
Sifnos	553,500	53.0	40.5	4.5	55.0
Limnos	1,827,900	46.3	39.8	5.8	54.4
Souli	1,287,740	40.9	34.6	5.4	60.4
Korinthos	5,906,214	40.2	32.6	5.3	62.1
Chios	4,229,898	38.9	31.3	4.5	64.1
Dorida	1,207,440	30.8	25.4	4.3	70.2
Trifylia	3,222,224	30.6	25.4	4.5	70.1
Fournoi Korseon	201,610	26.7	22.0	2.7	75.3
Mean (n = 11)	—	46.2	39.5	4.7	55.8

Note: Percentages are those reported or derived from the provider water balances. Totals may not sum to exactly 100 because of rounding and balance components that the submitted balances aggregate. Bold marks the summary row of unweighted sample means.

; Figure 3a). Real losses ranged from 22.0% to 67.5% of SIV. Apparent losses occupied a narrower reported range of 2.7–7.6%. Because apparent losses were relatively uniform in the submitted balances, the NRW ordering closely followed the reported real-loss shares. This is a property of the sample and its estimation procedures, not evidence that apparent losses are truly uniform among Greek providers.

The six island providers had a mean NRW of 49.1%, compared with 42.7% for the five mainland providers. Within-group ranges were much larger than the difference between means: 26.7–66.5% for islands and 30.6–71.1% for mainland providers. The sample is too small and non-random for an island effect to be inferred; broader studies likewise show that structural conditions explain only part of cross-utility variation [34,41].

Table 3. Reported non-revenue water and leakage levels for 2024 and 2025, from the Table B2 regulatory reference tables.

Provider	NRW 2024 (%)	NRW 2025 (%)	Change (pp)	Reported Leakage 2024 (%)	Reported Leakage 2025 (%)	Change (pp)	Balance Real Losses Minus Reported Leakage, 2025 (pp)	Agreement Between the Reported Leakage Fields
Aktio-Vonitsa	70.1	70.1	0.0	67.5	67.5	0.0	0.0	Numerically aligned
Leros	68.0	65.0	−3.0	62.9	60.5	−2.4	0.0	Numerically aligned
Alonnisos	62.7	63.3	+0.6	54.2	54.7	+0.6	0.0	Numerically aligned
Sifnos	53.0	53.0	0.0	40.5	40.5	0.0	0.0	Numerically aligned
Limnos	40.0	39.8	−0.2	22.0	22.3	+0.3	+17.5	Unreconciled discrepancy (Section 3.5)
Chios	39.5	39.5	0.0	22.0	22.3	+0.3	+9.0	Unreconciled discrepancy (Section 3.5)
Souli	39.0	39.0	0.0	24.9	25.0	+0.1	+9.6	Unreconciled discrepancy (Section 3.5)
Korinthos	42.5	36.1	−6.5	34.8	28.8	−6.0	+3.8	Unreconciled discrepancy (Section 3.5)
Trifylia	30.8	29.9	−0.9	26.2	25.4	−0.8	0.0	Numerically aligned
Dorida	29.8	29.8	0.0	25.4	25.4	0.0	−0.1	Numerically aligned
Fournoi Korseon	26.7	26.7	0.0	22.0	22.0	0.0	0.0	Numerically aligned

Note: NRW is field D.1.4 (non-revenue water of the internal distribution network), and the reported leakage level is field D.2.1 of each provider’s Table B2. These values are not directly comparable with those of Table 2, which derive from the submitted water balance for the system as a whole. The difference between the two, 70.1% and 71.1% for Aktio-Vonitsa, for example, is consistent with a difference in reporting boundary, but the dossiers do not state the boundary of either field, so the difference is reported here without attributing a cause. The penultimate column is the difference between the real losses derived from the 2025 water balance (Table 2) and the leakage level reported for the same year; a positive value means the balance reports more real losses than the reference table. The final column records whether the two figures agree numerically: the seven providers marked as numerically aligned differ by 0.1 percentage points or less, and the four marked as unreconciled differ by 3.8 to 17.5 points. Numerical agreement is a check applied in this study; it does not establish that the two fields share a definition or that the rest of the dossier is validated. Year-on-year changes are computed from the unrounded values recorded in the dossiers and may therefore differ in the last digit from the difference in the rounded figures displayed: for Korinthos, for example, the reference table records 42.54% for 2024 and 36.06% for 2025, a change of −6.48 percentage points, shown here as −6.5.

sets the non-revenue water and the leakage level that the reference tables record for 2024 and 2025 side by side; Section 3.5 returns to the agreement between those two fields.

3.2. Balance Composition and Metering Exposure

The balance composition shows two distinct billing structures (Figure 4). Fournoi Korseon and Trifylia reported billed-unmetered consumption equal to 36.6% and 31.6% of SIV, or 50.0% and 45.5% of their billed consumption. The remaining providers were almost fully metered in the submitted balance. Billed-unmetered consumption is legitimately classified as revenue water. The comparability issue is evidential: its volume is estimated rather than registered, and any estimation error transfers one-for-one to NRW [12,23,42,43,44].

A 10%, 20%, or 30% error in the billed-unmetered estimate would change NRW by 3.7, 7.3, or 11.0 percentage points in Fournoi Korseon and by 3.2, 6.3, or 9.5 points in Trifylia (

Table 4. Metering coverage and sensitivity of NRW to error in billed-unmetered consumption.

Provider	Billed Unmetered (% SIV)	Metering Coverage of Billing (%)	Reported NRW (%)	10% Error (pp)	20% Error (pp)	30% Error (pp)	Maximum Exposure (pp)
Fournoi Korseon	36.6	50.0	26.7	±3.7	±7.3	±11.0	36.6
Trifylia	31.6	54.5	30.6	±3.2	±6.3	±9.5	31.6
Chios	0.7	98.9	38.9	±0.1	±0.1	±0.2	0.7
Sifnos	0.7	98.5	53.0	±0.1	±0.1	±0.2	0.7
Leros	0.5	98.5	66.5	±0.1	±0.1	±0.1	0.5
Other six providers	0.0	100.0	30.8–71.1	0.0	0.0	0.0	0.0

Note: The 10–30% columns are symmetric scenarios for error in the billed-unmetered estimate. Maximum exposure is the full billed-unmetered share and is a one-sided structural counterfactual, not a confidence interval.

; Figure 5). The full “metered-only” case is therefore used only as maximum structural exposure: it shows how much of reported performance rests on the unmetered estimate, not the likely error. No evidence in the dossiers establishes the direction or magnitude of the actual error.

3.3. The Reported ILI Cases

Table 5. Infrastructure and UARL inputs for the seven ILI cases.

Provider	Year (Source)	Lm (km)	Nc	P (m)	λp (m/connection)	T (h/day)	Density (conn./km)	Nc + 20Lm
Aktio-Vonitsa	2024, 2025 (B2)	200.0 a	10,000 a	35.0 b	4.0	24	50.0	14,000
Leros	2024, 2025 (B2)	60.1 c,d	4313	40.0 e	≈0	24	71.8	5515
Alonnisos	2024, 2025 (B2)	26.3 c	2000 a	40.0 e	≈0	24	76.0	2526
Fournoi Korseon	2024, 2025 (B2)	5.4 d	1000 a	40.0 e	≈0	24	185.2	1108
Sifnos	2024, 2025 (B2)	177.0 f,g	3675	38.8 g	15.0 g	24	20.8 g	7215
Chios	2023 (WB-EasyCalc)	2000.0 f	35,456	38.8	15.0	4	17.7	75,456
Souli	2025 (WB-EasyCalc)	470.0 f	7019	60.0 h	15.0	24	14.9	16,419

Note: Inputs are those adopted in each ILI case, read from the leakage-assessment report held in sub-folder 06 of the dossier or from the WB-EasyCalc file, alongside the Table B2 reference table. Every value in the P and Lp columns is stated in one of those documents; none is documented as a measurement. The final column evaluates the 2009 minimum-size expression Nc + 20Lm > 3000; Alonnisos (2526) and Fournoi Korseon (1108) do not satisfy it. λp is the mean private-pipe length per connection; Lp in Equation (4) is the corresponding total in kilometres, that is, λp multiplied by Nc. ^a Reported in the dossier as an estimate or as a round figure. ^b The report states a range of 30–40 m; the mid-point is shown. ^c Value recovered from the reported annual real losses per kilometre; it agrees with the distribution-network length stated directly in the leakage report and in Annex II. ^d The length adopted is the distribution network alone. ^e The same round value of 40 m is adopted in three independent reports. ^f The length adopted is the sum of the transmission and distribution networks reported in Annex II. ^g For Sifnos, the row shows the basis behind the ILI reported in the reference table; the leakage reports in the same dossier use 115 km, a private-pipe length of zero, a pressure of 38 m, and a density of 32.0 connections per kilometre, as Section 3.5 sets out. For Alonnisos, the transmission and distribution conventions coincide because Annex II reports no transmission main; for Aktio-Vonitsa, no breakdown is reported. ^h Stated in the source and marked there as unverified.

reports the denominator inputs for the seven ILI cases and evaluates the 2009 size expression Nc + 20Lm > 3000 [13]. Five of the seven cases satisfy it; Alonnisos (2526) and Fournoi Korseon (1108) do not, which places their standard UARL outside the range for which the equation was calibrated and indicates that a system correction factor should be considered. Low connection density is a different matter: Souli (14.9 connections/km) and Chios (17.7) fall below the conventional threshold and Sifnos (20.8) sits just above it, but density governs the choice of volumetric unit rather than the validity of ILI. The material questions are whether the network and connection counts are complete, whether pressure is documented and representative, and whether CARL and time under pressure are comparable. Average pressure is stated for all seven cases once the leakage reports are read, and is documented as a measured, zone-weighted value in none of them; what the reports say about it is more revealing than its absence would have been. Three of them—Leros, Alonnisos and Fournoi Korseon—adopt the same round figure of 40 m, and the Fournoi Korseon report states in terms that the value is an assumption of the analysis rather than an observation. The Sifnos report records that no systematic measurement of pressure by zone exists. Aktio-Vonitsa states a range of 30 to 40 m, of which the midpoint is used, and Souli states 60 m and marks it as unverified. Because ILI varies inversely with pressure, every index in this set is conditional on a stated but unmeasured input. The recurrence of one round value in three submissions is a repeated stated assumption rather than an observation; no inference is drawn here about why it recurs, and the dossiers concerned were not all prepared independently of one another, as the Conflicts of Interest statement records. Table 5 records the pressure and the private-pipe length adopted in each case on that footing. Two of the mains lengths in Table 5 are shown as recovered quantities. For Leros and Alonnisos, the leakage report expresses real losses per kilometre of mains, and dividing the annual loss volume by that figure returns 60.1 and 26.3 km. Both agree with the distribution-network length that the same report and Annex II state directly, 60 and 26.3 km, so the recovery corroborates the reported length rather than substituting for it. It does not make the per-kilometre indicator an independent check: qL computed from a length recovered in this way reproduces the reported unit loss instead of testing it, and for these two providers qL in

Table 6. Reported and derived leakage indicators for each ILI case-year.

Provider	Year	CARL (m3/day) a	UARL (m3/day)	ILI b	qRL (L/conn./day w.s.p.)	qL (m3/km/day w.s.p.)	Principal Caveat
Aktio-Vonitsa	2024	9252	441	21.00	925	46.26	Mains length and connection count reported as estimates; pressure as a stated range
Aktio-Vonitsa	2025	9422	440	21.40	942	47.11	Mains length and connection count reported as estimates; pressure as a stated range
Leros	2024	2436	182	13.40	565	40.53 c	Pressure as a round 40 m assumption; mains length recovered
Leros	2025	2353	181	13.00	546	39.15 c	Pressure as a round 40 m assumption; mains length recovered
Alonnisos	2024	475	83	5.70	238	18.06 c	Below the size expression; pressure as a round 40 m assumption
Alonnisos	2025	541	83	6.50	271	20.58 c	Below the size expression; pressure as a round 40 m assumption
Fournoi Korseon	2024	122	36	3.39	122	22.50	Below the size expression; pressure stated as an assumption; identical reported values in both years
Fournoi Korseon	2025	122	36	3.39	122	22.50	Below the size expression; pressure stated as an assumption; identical reported values in both years
Sifnos	2024	655	291	2.25	178	3.70	Reference table and leakage report of the same dossier use different inputs (Section 3.5)
Sifnos	2025	614	291	2.11	167	3.47	Reference table and leakage report of the same dossier use different inputs (Section 3.5)
Chios	2023	5567	503 (3019 at 24 h)	11.06	942	16.70	Intermittent supply: eight hours on 3.5 days a week
Souli	2025	1249	1002	1.25	178	2.66	Pressure assumed and explicitly unverified; CARL not reconciled with the reported leakage level

Note: qRL and qL are recalculated from the displayed CARL, Nc, Lm and T; for Aktio-Vonitsa and Sifnos they reproduce the values printed in the dossiers (925 and 942 L/connection/day, and 178.3 and 167.1 L/connection/day, respectively). UARL is the value the provider’s own leakage report computes from the inputs of Table 5, and equals CARL divided by the reported ILI; it is not an independent calculation. Chios’s UARL of 503 m³/day is the 24 h value of 3019 m³/day scaled by the four hours for which the system is pressurised, so CARL and UARL are both expressed per calendar day. Alonnisos and Fournoi Korseon fall below the 2009 minimum-size expression (2526 and 1108 against a threshold of 3000). ^a CARL is derived from the top-down water balance and is an estimate of real losses rather than a measurement. ^b No case rests on a pressure documented as measured, so every ILI in this table is conditional on the pressure input recorded in Table 5. ^c Computed from a mains length recovered from the reported annual real losses per kilometre; the value reproduces the reported figure and is not an independent recalculation.

should be read as a restatement of the submission. Their reported ILI is unaffected, because the UARL shown in Table 6 is the value their own reports compute from the stated inputs and was not recalculated here.

The twelve reported values span an ILI of 1.25–21.40, a qRL of approximately 122–942 L/connection/day w.s.p. and a qL of 2.7–47.1 m³/km/day w.s.p. (Table 6; Figure 6). These ranges are descriptive. The apparent association between ILI and qRL is partly mechanical because both contain CARL, and ILI can be written as qRL divided by the per-connection UARL normaliser. It should not be presented as independent evidence that one measure validates or replaces the other. Five providers contribute two records each. These are retained in Table 6 for transparency about the reporting cycles from which the values come, and are not interpreted here as documented changes in performance; the material does not separate a change in the network from a change in the reported figure.

Aktio-Vonitsa, Chios and Souli illustrate three different interpretation problems. Aktio-Vonitsa reports the highest values in the sample, an ILI of 21.4 and 942 L/connection/day w.s.p. in 2025, but both its mains length (200 km) and its connection count (10,000) are reported estimates, and the dossier itself gives a scenario range of 17.4–26.8. Chios combines a 2023 ILI of 11.06 with a supply regime of eight hours on three and a half days a week, an average of four hours a day, so its time-normalised loss indicators are not directly comparable with continuously supplied systems without an explicit regime label. Souli has a 2025 ILI of 1.25 under an assumed pressure of 60 m; because that pressure has not been confirmed, the value is a scenario result rather than a measurement. The contrast between the highest and the lowest value in the sample therefore cannot be attributed to network condition alone, because these cases differ in the completeness of their documentation as well as in their reported losses; the relative contribution of the two has not been quantified here.

3.4. Sensitivity of the ILI Denominator

Normalising UARL per connection gives (18/d + 0.8 + 25λp)P, where d is connection density and λp is private-pipe length per connection in kilometres. Figure 7 shows the expected consequence: a sparse system has more mains per connection and therefore a larger per-connection technical reference. This is the purpose of normalisation, not evidence that ILI “breaks”. The seven cases span a per-connection normaliser of 36–143 L/connection/day, and Figure 7 draws the technical reference separately for each of the three private-pipe conventions in use, at the 40 m that three of the cases adopt. A case whose stated pressure differs from 40 m sits off its curve in proportion to that difference, so the vertical position of a case combines density, pressure and the private-pipe assumption; none of the three is a measurement, and the figure should not be read as a check on any of them. The practical vulnerability appears when d is wrong because mains length is incomplete, estimated or defined on a different basis, as in Aktio-Vonitsa, Sifnos and Souli, or when P is a weakly supported system-wide figure, as in Leros, Alonnisos and Fournoi Korseon, where three independent reports adopt the same round 40 m.

For Souli, keeping CARL and all non-pressure inputs fixed gives ILI values of 2.14 at 35 m, 1.66 at 45 m, and 1.25 at 60 m (Figure 8a). This range does not identify the correct ILI; it shows why the 60 m input must be verified. The one-at-a-time perturbation confirms direct inverse sensitivity to pressure, which alone moves the index by −16.7% and +25.0% for a ±20% change, and smaller, case-specific sensitivity to mains length (−9.2% and +11.3%), connection count (−6.3% and +7.2%), and private-pipe length (−3.0% and +3.2%) (Figure 8b). A defensible interpretation therefore requires a measured or hydraulically weighted average pressure and confidence limits for CARL [13,38,39].

3.5. Data Adequacy and Limits of the Submitted Balances

The data-adequacy screen (Figure 9) replaces a pass/fail matrix based on obsolete fixed thresholds. No case is free of documentation gaps. Sifnos, which on the reference table alone looks the most complete case in the set, is in fact the clearest illustration of the problem. Its reference table reports an ILI of 2.25 for 2024 and 2.11 for 2025, on a UARL of 106,273 m³/year built from 177 km of mains, 15 m of private pipe per connection and a pressure of 38.8 m. The leakage reports held in sub-folder 06 of the same dossier, for the same two years, use 115 km, a private-pipe length taken as zero, and a pressure of 38 m; they obtain a UARL of 69,489 m³/year and, for 2025, an ILI of 3.23 under their base loss scenario and 3.69 under their alternative. Connection density follows the same split: 20.8 connections per kilometre on the reference-table basis and 32.0 in the report. Neither calculation is arithmetically wrong. The dossier contains two internally consistent calculations that never meet, and a benchmark reading the reference-table field would place this provider in a different IWA band from one reading the report. Every other case carries at least one issue that materially affects comparison: mains length and connection count reported as estimates in Aktio-Vonitsa; a round 40 m pressure with private-pipe length taken as zero in Leros, Alonnisos and Fournoi Korseon; values repeated unchanged between reference years in Fournoi Korseon; intermittent operation in Chios; and an unverified pressure together with an unreconciled CARL in Souli. Private-pipe length is stated in all seven cases but measured in none: three conventions are in use across the set—zero, 4 m per connection, and 15 m per connection—and the choice is nowhere justified by measurement. These flags call for verification or scenario reporting; they do not prove that the underlying networks perform well or poorly.

Three further patterns limit the register’s analytical resolution. The first concerns internal consistency. For seven of the eleven providers, the leakage level reported in field D.2.1 of the reference table matches the real losses in the submitted water balance to within 0.1 percentage points, but for four it does not: the gap is 17.5 points for Limnos, 9.6 for Souli, 9.0 for Chios and 3.8 for Korinthos (Table 3; Figure 10a). In Souli, the same divergence appears in absolute terms, because the WB-EasyCalc CARL of 1249 m³/day implies 455,885 m³/year, or 35.4% of system input, against the 25.0% reported in the reference table. Two fields of the same dossier therefore describe different quantities of leakage, and a benchmark assembled from either field alone would rank these providers differently. The submitted material does not identify the cause, and at least three explanations remain consistent with it: a transcription or reporting error in one of the two entries; a difference in definition or reporting boundary, since field D.2.1 refers to the internal distribution network while the balance covers the system as submitted; and a different calculation baseline, data version, or reference period behind the two figures. The dossiers do not state which applies, and distinguishing between them would require the providers’ underlying calculation sheets, which are not part of the submission. Two of the four affected providers, Limnos and Korinthos, contribute no ILI case, so the inconsistency is a property of the register rather than of the ILI subset. In Souli, the reconciliation is three-way rather than two-way: the balance reports real losses of 34.6% of system input, the reference table reports 25.0%, and the WB-EasyCalc CARL corresponds to 35.4%. None of the three has been adjusted here to agree with the others.

The second concerns the composition of the loss estimate: nine of the eleven 2025 balances used the same 72.2/5.6/22.2 split of real losses among mains, storage and service connections (Figure 10b). That fixed allocation is a bookkeeping assumption, not evidence about where leakage occurs. Intervention design would require district metering, minimum-night-flow analysis or other field evidence [39,40]. Because the same allocation is submitted by nine providers whose networks differ in age, material and topography, it cannot support the targeting of mains replacement, storage rehabilitation or connection renewal, and it should not be used to justify investment in any of them.

A third pattern concerns the definition of mains length itself. Annex II of each dossier reports the transmission and the distribution network separately. For Souli and Chios, the WB-EasyCalc file uses the sum of the two, 470.0 and 2000.0 km against 190 + 280 and 1000 + 1000 km, and the same sum of 177.0 km underlies the ILI that the Sifnos reference table reports—although the Sifnos leakage report itself uses the distribution network alone, as described above. For Leros and Fournoi Korseon, the length is the distribution network alone: 60.1 and 5.4 km against reported totals of 120 and 9.4 km. Neither convention is stated in the submissions, and neither can be supported except where the length corresponds to the same system boundary as CARL and the other inputs; that correspondence is nowhere documented. The two are not interchangeable: the choice moves connection density, the size expression and the per-kilometre loss indicator, and it is invisible in a table that reports a single figure for mains length. The values in Table 5 are reproduced as the source used them, and the convention is recorded there so that the comparison is not made silently.

3.6. Proposed Reporting Scheme

Table 7. Proposed tiered reporting scheme for small and medium water systems.

Tier	Indicators	Minimum Documentation	Primary Use	Independent Precedent for the Requirement
1. Water balance	NRW (% SIV and m3/year); real and apparent losses; billing-metre coverage	Validated balance; a single stated source field per indicator; uncertainty notes	Resource, revenue and supervisory overview	IWA indicator family [3,4]; AWWA M36 water-audit data-validity grading [5]; ARERA conditions its loss indicator on metering and process quality [26]
2. Volumetric backbone	Real losses in L/connection/day and m3/km/day, both w.s.p.	Connections, complete mains length, hours pressurised	Benchmarking by system type; operational targets	Ofwat publishes leakage in absolute and normalised volumetric form [27]; ARERA reports linear losses in m3/km/day [26]; normalised volumetric indicators shown to track ILI closely [45]
3. Supplementary ILI	ILI plus CARL, Lm, Nc, Lp, P, T and calculation year	Audited inputs; documented pressure and supply regime; size expression Nc + 20Lm evaluated; SCF considered where appropriate	Technical interpretation and diagnostic comparison	Directive (EU) 2020/2184 permits ILI or another appropriate method [25]; UARL guidance requires the inputs and a size check [13,14,15]; screening of the underlying measurement shown to govern ILI credibility [46]

Note: w.s.p. = when system pressurised. The scheme does not prescribe one universal ranking; the indicator should match the regulatory question. The final column identifies established practice that supports each requirement independently of the present sample; that precedent, together with the algebra of the indicators, is the conceptual basis of the scheme and not a demonstration that it works in practice. The scheme is proposed rather than validated: it has not been applied to an independent set of providers, the prerequisites and the order in which the tiers can be reached, together with the testing and costing that remain outstanding, are set out in Section 4.5.

translates the findings into a tiered reporting scheme. Tier 1 retains NRW as both percentage and annual volume and adds billing-metre coverage. Tier 2 reports real losses in both L/connection/day and m³/km/day when the system is pressurised; the preferred comparator can then be selected according to network density and management purpose. Tier 3 reports ILI only with the UARL inputs, supply regime, calculation year and an account of data quality. Most of the inputs the scheme requires already exist somewhere in the dossier, and reaching Tier 1 is therefore partly a matter of identifying the source field for each indicator. Only partly: where two documents of the same dossier give different values, as Section 3.5 shows they can, the fields have first to be reconciled. Tier 2 asks more again, since several of the mains lengths used here are estimates or recovered quantities and would need asset records that can be audited rather than cited. Average pressure is stated for all seven ILI cases and documented as a measurement in none, so Tier 3 would require new monitoring; the cost of that monitoring for providers of this size has not been assessed here. The final column of Table 7 records the independent precedent for each element of the scheme, so that no tier rests on this sample alone.

4. Discussion

4.1. Indicator Disagreement Should Not Be Overstated

The material examined does not support a claim that NRW and ILI produce an almost inverted ranking. The two indicators are documented to different depths in the same dossiers, and the cases with the most extreme ILI values are also the cases with the weakest denominator evidence. What the data do support is narrower and more useful: NRW%, volumetric losses, and ILI encode different denominators; their values can diverge for understandable reasons; and data quality can dominate interpretation in a small benchmark. This framing is consistent with the IWA indicator-family approach and with recent critiques of using any single loss indicator for asset-management targets [3,9,47].

NRW% is informative about the share of input volume that does not generate revenue, but it does not isolate infrastructure condition. High consumption can reduce the percentage for an unchanged loss volume, while low consumption can increase it. ILI normalises CARL to asset exposure and pressure, but its interpretability depends on the credibility of those inputs. The per-connection and per-kilometre indicators are transparent volumetric measures, yet they also answer different questions and are influenced by density. No single metric is denominator-free. The reconciliation gaps documented here add a further caution: before indicators are compared at all, the fields from which they are read must be shown to describe the same quantity.

4.2. Metering Coverage Is Part of Indicator Quality

The strongest empirical finding concerns billed-unmetered consumption. For two providers, a large share of apparent performance depends on an estimated revenue-water component. The ±10–30% scenarios avoid presenting the full metered-only counterfactual as a statistical uncertainty band. They show that even moderate proportional error could shift NRW by several percentage points. The correct regulatory response is not to reclassify lawful billed-unmetered consumption as loss, but to publish billing-metre coverage and the estimation method alongside NRW [5,12,23,42,43,44].

This requirement would also improve international comparability. ARERA’s technical-quality framework conditions its loss indicator on process and customer-metre coverage, illustrating that an indicator can be paired with data-quality prerequisites [26]. For the Greek framework, the necessary first step is a reproducible audit trail: source volume, estimation rule, population or connection basis, and uncertainty range. Full volumetric metering may be a long-term objective, but the present data do not establish its economic feasibility for every small provider.

4.3. ILI in Small and Intermittently Supplied Systems

A rule-based interpretation can treat 3000 connections, 20 connections/km, a 25–100 m pressure range, and continuous supply as four universal applicability criteria. That interpretation is not supported by the updated history of UARL guidance. The density restriction was removed, the size criterion evolved, and SCF methods were developed for small systems and pressure/pipe combinations where the standard equation may misestimate the technical reference [13,14,15]. Evidence from small supplies nevertheless shows why input quality and uncertainty remain important [48]. The present sample makes the distinction concrete. Souli and Chios fall below 20 connections/km, and Sifnos is marginally above it, yet all three satisfy the size expression comfortably, so their ILI can be computed and reported, provided the volumetric indicators accompany it. Alonnisos and Fournoi Korseon are the opposite case: their densities are high, but Nc + 20Lm is 2526 and 1108, below the threshold at which the standard UARL was calibrated. For those two systems, an SCF-adjusted reference, or reporting restricted to the volumetric indicators, is the more defensible option.

Intermittent supply requires separate treatment. Chios’s four-hour regime changes both qRL and qL by a factor of six relative to a calendar-day expression and reduces UARL in proportion to time under pressure. Repeated filling and transient flows also affect water-balance and customer-metre error [20,21,22,23,24]. The numerical ILI can be reported, but comparisons must retain the w.s.p. basis, supply duration, and method used to estimate CARL. A continuously supplied utility should not be compared with Chios through an unlabelled single number.

A further qualification applies to every CARL value used here. All are derived from top-down annual water balances, in which real losses are the residual left after authorised consumption and estimated apparent losses have been subtracted from system input. That residual absorbs error from every term above it: bulk-metre bias, customer-metre under-registration, the estimate used for billed-unmetered consumption and the assumed apparent-loss share [10,11,12,23,42,43,44]. Two features of this material make the absorption consequential. Where apparent losses are entered as a uniform assumption rather than assessed, as they are in most of these balances, any under-statement of apparent losses transfers one-for-one into real losses and inflates both the ILI and the volumetric indicators. Where supply is intermittent, metre under-registration at low and unsteady flows, air passing through metres on refilling, and losses during pipe filling all bias the same residual, and the direction of the net bias cannot be determined from the submitted material [20,21,22,24]. Under suitable conditions, a bottom-up minimum-night-flow analysis would provide an estimate of real losses that is independent of the balance residual and would therefore constrain it, although it would not by itself resolve where the losses occur; no dossier contains one [39,40]. The CARL values in Table 6 are therefore reported as water-balance-derived estimates rather than as measured leakage, and the ILI values that rest on them inherit that status. The term real losses is retained in its standard IWA sense and does not imply direct measurement.

4.4. Comparison with Published Work

Placing this material beside published work helps to separate what is ordinary from what is specific to these dossiers. Two comparisons concern the values themselves. The 2025 non-revenue water range of 26.7–71.1% is wide but not exceptional: European evidence assembled across countries and utilities reports comparable dispersion and shows that national averages conceal much larger utility-level variation [2], and earlier Greek work reports urban and island losses of a similar order [9,32,33,34]. Determinants studies in comparable European settings find that structural characteristics explain only part of the variation between utilities, which is consistent with the within-group ranges reported in Section 3.1 exceeding the difference between the island and mainland means [41]. Work on small supplies likewise reports that indicator values in systems of this size are strongly conditioned by the completeness of the underlying records [48]. Beyond these qualitative points, we do not compare our ILI values numerically with published ones: they rest on different time bases, different supply regimes and inputs of differing provenance, and a numerical comparison would require the same information to be stated on both sides.

Three comparisons concern method, and they locate the present contribution more precisely. Work that examines how far the ILI can be made reliable has moved towards high-resolution metering: a recent single-DMA study uses advanced metering infrastructure to filter non-representative periods before an annual ILI is reported and concludes that credibility depends on the aggregation and screening of the underlying measurements rather than on the index formula [46]. That study improves an indicator where measurement exists; the present study asks what the indicator can carry where measurement does not exist, and the two are complementary rather than competing. Second, an assessment of Polish systems undertaken in view of Directive (EU) 2020/2184 reports that normalised volumetric indicators correlate closely with the ILI and may substitute for it where its calculation is problematic [45]. That study distinguishes basic normalised indicators, which contain no pressure term, from intermediate ones that are normalised by pressure, and the closest relations it reports are for the intermediate group. Tier 2 as proposed here corresponds to the basic group: it substitutes for the ILI as a comparator of loss volume and not as a pressure-normalised technical reference, which is why Tier 3 is retained separately. The reported relations were in any case obtained on systems whose inputs were available. Third, the critique of using a single loss indicator to drive asset-management decisions [47] is reinforced by this material, since the cases with the most extreme index values are also the cases with the weakest denominator evidence. What is less commonly available in this literature, and is available here, is two consecutive regulatory years for the same providers, which identifies repeated reported values that require clarification and that a single-year benchmark cannot detect, and a reported leakage level alongside the water balance in one submission, which makes internal consistency testable. Studies drawing on a single reported field can perform neither check. This is consistent with a benchmarking literature that has long held indicator comparison to be only as strong as the validation of the underlying data [36,39].

4.5. Regulatory Implications

The proposed hierarchy is compatible with Directive (EU) 2020/2184 because the Directive permits ILI or another appropriate method [25]. It also follows current practice in which regulators publish complementary measures rather than a single percentage [26,27]. Percentage NRW remains useful for communication and resource accounting; absolute NRW links performance to water and financial consequences; qRL and qL provide transparent physical normalisations; and ILI adds a technical reference when inputs are auditable. Reporting annual volumes also supports the resource-efficiency and cost-recovery context of the Water Framework Directive [49].

Target setting should occur within comparable system types and should consider the economic level of leakage, service standards, scarcity and intervention cost [50]. A low ILI does not prove that further reduction is uneconomic, and a high NRW percentage does not by itself identify the most cost-effective project. The dossier can support screening, but funding decisions require verified measurements, pressure information, and an intervention appraisal. The companion carbon analysis addresses a separate consequence of the same losses [51].

The conditions under which the proposed scheme would transfer should be stated, since they are not established by this sample. The scheme is written for small and medium providers reporting under a common regulatory template, and it assumes that a reconciled water balance is prepared, that connections and mains length are held in a register that can be audited, and that the reporting authority can specify the field from which each indicator is read. Where those conditions hold, nothing in the scheme depends on the size or the location of the providers studied here. Where they do not, the tiers describe an order of construction rather than an order of availability: Tier 1 is then the objective of assembling and reconciling the basic data, not a level of reporting already in place, and Tiers 2 and 3 require the corresponding measurement before they can be reached. The reporting effort should not be understated either. Identifying the source field for each indicator is largely a documentation task, but the reconstructed lengths and unreconciled volumes documented in Section 3.5 show that reaching Tier 2 will in several cases require asset records to be assembled rather than merely cited. Two questions are left open. The scheme has not been applied to an independent set of providers, so its effect on comparability is proposed rather than demonstrated; and the cost of the monitoring implied by Tier 3, pressure logging in particular, has not been appraised for providers of this size.

4.6. Limitations

Five limitations bound the results. First, the benchmark contains eleven purposively selected providers and is not representative of the Greek sector. Second, the ILI analysis covers seven of those eleven providers; five contribute both reference years from the regulatory reference table, while Chios contributes a 2023 case and Souli a 2025 case from WB-EasyCalc, so the twelve values are not a balanced panel. Third, CARL is based on top-down balances rather than independent minimum-night-flow estimates, and is reported throughout as a water-balance-derived estimate rather than as measured leakage. Fourth, several infrastructure inputs are estimates or recovered quantities; private-pipe length is stated in every case but assumed in every case, under three different conventions, and the arithmetic recovery performed here reproduces those assumptions rather than testing them. Fifth, the metering analysis evaluates sensitivity to assumed error; it does not estimate actual billed-unmetered consumption error. These limitations preclude causal inference, national prevalence estimates, and provider league tables.

5. Conclusions

This study asked what the water balances of eleven purposively selected Greek providers show descriptively, how strongly metering practice can affect non-revenue water, how the reported ILI values should be read when their denominator inputs are documented to different degrees, and what reporting scheme is defensible with the information already collected. The answers concern the interpretability of indicators in the material examined; they are not statements about the performance of the Greek water sector.

For non-revenue water, interpretation was found to depend on two things that the percentage itself does not display. The first is how much of billed consumption is estimated rather than registered: in two providers, billed-unmetered consumption approached a third of system input, so an error of 20% in that single estimate would move the reported figure by more than six percentage points. The second is which field the value is read from. Six of the eleven providers reported an identical figure in both reference years, a pattern no single-year benchmark could reveal; in four dossiers, the reported leakage level and the real losses in the same provider’s balance differ by between 3.8 and 17.5 percentage points, and in one the reference table and the leakage report of the same year compute the ILI from different inputs, so the source document must be specified before any comparison is attempted.

For the ILI, the conclusion is one of readability rather than of ranking. The twelve values span 1.25 to 21.40, but they are documented to different depths: mains length is an estimate or a recovered quantity in three cases and follows two different conventions across the set; private-pipe length is assumed in all seven under three different conventions; one provider repeats its inputs unchanged between years, one operates an average of four hours a day, and in one dossier the reference table and the leakage report of the same year yield different indices from different inputs. Two systems fall below the 2009 minimum-size expression and would be better served by a system correction factor or by the volumetric indicators alone, whereas connection density below 20 connections per kilometre is a question of which volumetric unit to report and not a reason to reject the standard equation.

What follows for reporting is a documented indicator set rather than a single headline figure: non-revenue water in percentage and volume, billing-metre coverage, real losses per connection and per kilometre when the system is pressurised, and ILI accompanied by its denominator inputs, reference year, supply regime and the provenance of each input. The conceptual basis of that scheme is the algebra of the indicators together with existing regulatory precedent, not this sample; the sample shows which of its safeguards address failure modes that are present in submitted material. The scheme has not been tested on an independent set of providers.

Three limitations bound these conclusions, and the first is the most consequential. No ILI case in this sample rests on a documented pressure measurement: average pressure is stated for all seven cases and documented as a measurement in none; three of the seven adopt the same round value of 40 m, and one dossier marks its own figure as unverified. Because ILI varies inversely with pressure, every value reported here is conditional on an input that the dossier does not document as measured, and the absence of documented, hydraulically weighted pressure is the primary obstacle to interpreting the ILI data. Second, all CARL values are water-balance-derived residuals rather than independent minimum-night-flow estimates and therefore absorb the error of every term above them. Third, the sample is purposive, and the twelve ILI values form a case set with incomplete information rather than a balanced panel, which precludes prevalence estimates, causal inference, and provider league tables.

Three lines of work follow directly. Zone-weighted pressure monitoring in a small number of representative providers would remove one substantial source of uncertainty from the ILI; the index would remain a computed quantity, and a defensible statement of its uncertainty would still require the remaining inputs—CARL, mains length, connections, private-pipe length and hours under pressure—to be documented and their uncertainties propagated. Minimum-night-flow analysis in the same systems would, under suitable conditions, constrain CARL independently of the balance residual, although establishing where losses occur, and therefore testing the uniform loss-component allocation that nine of these providers submitted, would require additional targeted measurement such as district metering. Extending the reference-table cross-check to the full national register would then establish how often the inconsistencies documented here occur, which is the prevalence question that the present design deliberately does not answer.