Open Water has published a final report on the results of a data pilot to see if external datasets could be used to help water companies identify non-household premises in their areas of supply, ahead of the opening of the competitive market in England in April 2017.

Eight companies which supply around 60% of non-household properties in England between them took part in the data pilot, spending nine months on the matching process. Participating companies were:

• Albion Water
• Anglian Water
• Cambridge Water
• Essex & Suffolk Water
• Northumbrian Water
• Severn Trent Water
• South Staffordshire Water
• Thames Water

The companies were chosen for a number of reasons, including operational areas bordering with Scotland and bordering/entering into Wales, premises served by more than one incumbent, incumbents covering two distinct geographies and challenges posed by large urban areas in terms of multi-occupancy premises.

The pilot companies had the freedom to choose their own matching methods  - there was no consistent approach to validation, which has implications for the comparability of results.

The pilot identified 1,663,762 non-household premises in England (each with a Unique Property Reference Number (UPRN)) from the National Address Gazetteer and the Valuation Office Agency records of properties liable for non-domestic rates. 1,092,790 of the properties were in the areas supplied by the eight pilot companies.

Headline results show:

• 68% of non-household premises matched to individual water company records;
• 31% of water company non-household records were not found in the list of non-household premises;
• 23% of non-household premises provided were not found in water company records; and
• there was a wide range of success in matching among the pilot companies - ranging from under 40% to over 95%.

Open Water said three factors appear to have led to the variation in success among the companies:

• data structure in the pilot company systems;
• historic quality of pilot company data; and
•  the use of different thresholds for what was sufficiently close to be an exact match.

Common challenges faced by the pilot companies included identifying exactly which company a premise was served by at company boundaries, ensuring consistent treatment of premises receiving water from one company and sewerage from another, extra work for domestic and commercial premises with similar addresses and not keeping records of premises not served.

Each of the pilot companies experienced different issues and achieved different results, with  techniques that worked well for some companies less successful for others. They also had data organised in different ways.  Open Water said this demonstrated the importance of companies using matching techniques tailored to the structure and quality of their own data.

Commenting on the implications of the trials for market preparedness, Open Water said:

“Every water company needs to analyse their current data and then have a clear plan to ensure they can give the market operator and Ofwat the necessary assurances that their non-household data is accurate and complete.”

“It is clear that while the automated and semi-automated approach provides a good start, it is not sufficient to provide a robust full match without rigorous validation.”

“Additional steps will be necessary to confirm that all non-household premises have been unambiguously identified for inclusion in the competitive market.”

“It is important that companies master the finalised Ofwat eligibility guidance and apply the guidance consistently to their records to identify non-household premises to include on the central market register. This is vital to ensure a level playing field for retailers and customers.”

The report also says it was never expected that the results of the pilot would be used to establish a process which companies had to follow.

Click here to download the final report.