Water & utilities data: smart metering, leakage and demand

Water utilities sit on enormous operational data of their own, yet they still need external data to put it in context. From smart-meter consumption to leakage indicators and demand forecasting, this guide covers what water and utilities data exists, why it matters under tightening EU rules, and how to source and deliver it reliably.

Why water data matters now

Europe’s water networks lose a striking share of treated water before it reaches a customer. According to EurEau, the average level of non-revenue water across European countries is around 25%, with national figures ranging from roughly 5% in the Netherlands to over 60% in Bulgaria. Every lost cubic metre carries energy, treatment and carbon cost, which is why leakage has become both an operational and a regulatory priority.

The recast EU Drinking Water Directive (Directive (EU) 2020/2184) brings leakage into the open: larger utilities, broadly those producing more than 10,000 m³ per day or serving more than 50,000 people, are required to assess and report water-loss levels to the European Commission, with EU-wide reporting due by early 2026 and a loss threshold expected to follow by 2028. Reliable data is now a compliance input, not just a management aid.

The water & utilities data landscape

• Smart-meter consumption: interval reads at household, district or grid-zone level.
• Leakage and non-revenue water indicators, including pressure and flow monitoring.
• Demand and consumption profiles, including irrigation and seasonal patterns.
• Network and asset data: pipes, valves, maintenance and renewal rates.
• Quality and environmental data: water and wastewater quality, reservoir levels, drought and flood-risk indicators.

Common use cases

Utilities and the organisations that serve them use this data to detect and prioritise leaks, forecast demand, plan asset renewal, model drought and flood risk, benchmark performance, and meet regulatory reporting obligations. Increasingly it also feeds digital-twin and AI models, which depend on consistent, well-structured historical and near-real-time inputs.

Sourcing considerations specific to water

Water data raises particular challenges. Metering intervals and definitions vary between operators; non-revenue water itself is measured with many different indicators, so comparability is not guaranteed. Personal data appears wherever consumption is tied to a household, so anonymisation or aggregation is often required. And much of the value lies in combining sources: meter data with weather, asset and topographic layers, which means harmonising formats and reference frames.

Delivery and cadence

Use cases span the full delivery spectrum. Asset-renewal planning may need a one-off historical extract; demand forecasting wants regular batches; leak detection and pressure management benefit from near-real-time or streaming telemetry over API or MQTT. The right model is defined around the utility’s operational systems, with formats such as Parquet or CSV for analytics and streaming interfaces for live monitoring.

Governance and privacy

Where consumption data can be linked to individuals, the GDPR applies and anonymisation, pseudonymisation or aggregation is typically needed. Provenance and licensing documentation matter especially for regulated reporting and tenders. Practices aligned with NIS2 and ISO/IEC 27001 principles are increasingly expected for critical-infrastructure data.

The systems behind water data

Understanding where water data originates helps you judge its quality and cadence. Most operational data flows from SCADA systems (the telemetry layer for pumps, valves, reservoirs and treatment works), from smart or AMR meters at the customer edge, and from district metered areas (DMAs), the discrete network zones that make it possible to localise losses. Around these sit asset-management systems, GIS network models, water-quality laboratory data and, increasingly, satellite and acoustic sources. A dataset that looks the same on a spec sheet can differ enormously depending on which of these layers it comes from and how it was sampled.

Turning leakage data into action

Leakage is the headline use case, and the data techniques behind it are well established. Minimum night flow analysis within a DMA isolates the period when legitimate demand is lowest, so persistent flow points to losses. Pressure management uses pressure and flow data together, because leakage rises with pressure, so trimming excess pressure cuts both losses and pipe stress. Acoustic and correlation data pinpoints leaks for repair. None of these works without consistent, well-timestamped data across the meter, DMA and SCADA layers, which is exactly where harmonisation effort pays off.

A worked example: assembling a demand dataset

Consider a utility that wants to forecast district-level demand. The requirement combines several external and internal layers: historical DMA inflow at 15-minute resolution, weather (temperature and rainfall) aligned to the same timestamps and geography, customer-mix and demographic context for each district, and calendar effects. The hard part is rarely any single source, it is aligning time zones and intervals, reconciling DMA boundaries with statistical geographies, and gap-filling sensibly. Done well, the result is a clean, documented dataset a forecasting model can consume directly; done badly, the model inherits every inconsistency.

Quality and acceptance criteria for water data

Because so much water data feeds regulated reporting, define acceptance criteria up front: completeness of interval reads above a stated threshold, freshness within the cadence the use case needs, valid value ranges (negative or impossible flows flagged), consistent units, and explicit handling of estimated versus measured reads. Tying these to the supply contract is what makes a recurring water feed trustworthy over time.

Sources & further reading

• EurEau, Europe’s Water in Figures (2021), non-revenue water statistics.
• EUR-Lex, Directive (EU) 2020/2184 (recast Drinking Water Directive).
• European Environment Agency, water resources and use indicators.
• EUR-Lex, Regulation (EU) 2016/679 (GDPR).