Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) - Version 1.00 | 17 February 2021
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 Authors Luke Austin luke.austin@mosl.co.uk Matt Labrum matt.labrum@mosl.co.uk
Contents 1. Introduction ..................................................................................................................................................... 2 1.1 Background ................................................................................................................................................ 2 1.2 Key highlights............................................................................................................................................. 3 1.3 Next steps .................................................................................................................................................. 5 2. Summary of RFI Responses.............................................................................................................................. 7 2.1 Impacts of poor data quality ..................................................................................................................... 7 2.2 Trading party activities to mitigate poor data quality ............................................................................. 12 2.3 Benefits of data quality ........................................................................................................................... 17 2.4 Paid for data services .............................................................................................................................. 19 2.5 Maintaining data quality ......................................................................................................................... 20 2.6 Data outside of CMOS ............................................................................................................................. 22 2.7 New technology and end-user verification ............................................................................................. 23 2.8 Customer details definition ..................................................................................................................... 24 2.9 Meter manufacturer validation ............................................................................................................... 25 2.10 Additional feedback............................................................................................................................... 26 3. Conclusion ..................................................................................................................................................... 29 Appendix I .......................................................................................................................................................... 30 Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 1
1. Introduction 1.1 Background As part of the Market Performance Operating Plan (MPOP) 2020/21 (see Figure 1, Workstream A: ‘High quality customer, premises and asset data’), MOSL committed to delivering a data cleanse plan for core market data items. Figure 1: Overview of MPOP for 2020/2021 We published a Request for Information (RFI) in October 2020 to inform the planning and prioritisation of the data cleanse activity. We sought feedback on: (i) the cost-impact and benefits of data quality; (ii) the required activities for mitigating or maintaining data quality and (iii) the recommended next steps for resolving data quality issues. The RFI consisted of ten questions covering the three core groups of data contained in the Central Market Operating System (CMOS): Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 2
Customer details and premises data Meter location data Meter details data. This document summarises the findings from the RFI and provides high level detail on the publication and delivery of the data cleanse plan. 1.2 Key highlights Overview The responses provided 97 per cent coverage of the market based on aggregate share of supply points, consisting of 14 wholesalers and 10 retailers Respondents welcomed the focus on addressing core market data issues and were broadly supportive of the approach; the focus on the key data items identified; and the proposals outlined to address data quality issues Some wholesalers questioned the value of ensuring completeness of unique property reference number (UPRN) and valuation office agency (VOA) reference data; however, the responses to the RFI have provided strong evidence that these premises data items are crucial for facilitating the fulfilment of core market obligations and activities The absence of occupancy status and Yearly Volume Estimates (YVE) from the scope of data cleanse was noted by some wholesalers; however, we can confirm that they are within scope for broader market improvement activity but have not been included here as they are already being addressed or investigated as part of other ongoing work. Impacts of poor data quality 80 per cent of retailers (eight) and 71 per cent of wholesalers (10) said they are adversely impacted by poor-quality customer and premises data, aside from the impacts already identified in MOSL’s analysis Based on the RFI responses, we estimate the minimum annual resource cost to the market from managing poor quality data to be £10m; with customer and premises details accounting for 39 per cent of this cost, meter location data 42 per cent and meter details data 19 per cent Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 3
Trading parties are deploying resource to obtain and verify missing or inaccurate data items, administer bilateral requests, manage consistency and mismatches between central and internal systems, resolve inaccurate or incomplete supply point data, address high volumes of failed meter reads and manage meter verification requests. Benefits of data quality Both retailers and wholesalers consistently highlighted that an improved customer experience was the main benefit of improved data quality through providing a better platform for excellent service, operational efficiency and meter reading services; thereby leading to greater customer satisfaction and trust, shorter customer journey times, lower performance penalties and improved accuracy of settlement charges, leakage calculations and customer billing. Paid for data services The majority of respondents said they are using third-party services to manage data or data quality issues, particularly for credit searches, verifying occupancy, verifying supply point eligibility, and obtaining or verifying customer and premises details. The types of services used include desk-based research, site visits, investigations, and data enrichment. Maintaining data quality Retailers are predominantly using manual processes to maintain data, but are increasingly moving towards automated processes, or aspire to do so, such as automated notifications following customer switches or new supply point registration Most wholesalers said they have proactive consistency checks to compare CMOS with their internal systems but rely on more reactive processes to resolve issues or anomalies detected through quality checks, site visits, customer complaints, bilateral requests, or periodic reviews. Data outside CMOS 70 per cent of retailers (seven) said there are differences between their internal systems and CMOS, primarily due to transactions failures, manual intervention, backlogs, system migrations and unintentional data enrichment 57 per cent of wholesalers (eight) said there are minimal differences between their systems and CMOS due to the automated processes in place to verify consistency; however, many said they hold Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 4
additional data that could be valuable to the market, such as photos of meter locations, meter location notes, logger identification and ownership details, more granular meter read data, non-billable premises and premises categorisation information. New technology and end-user verification 58 per cent of respondents (14) said they supported the use of new technology to capture end-user input and 42 per cent (10) were not supportive. Many of those that were not supportive said they would need more detail before agreeing, whereas others provided arguments against, including: it would weaken data ownership responsibilities; it would negatively impact a competitive element of the market or it would increase the cost of maintaining CMOS. Customer details definition The majority of trading parties said they interpreted ‘Customer Name’ (D2027) as the legal entity of the customer that is responsible for paying for the water consumption or services provided by the retailer; however, there was less consistency for the interpretation of ‘Customer Banner Name’ (D2050) field. In both cases there was support for providing greater clarity due to lack of consistency. Meter manufacturer validation 67 per cent of trading parties (16) supported removing meter manufacturer from the CMOS validation for meter read submissions, with many suggesting that it provides no benefit and hinders the submission of legitimate reads into the market. Of those that were not supportive, there were concerns that meters would not be uniquely identified in all cases and they argued that meter manufacturer data item should instead be standardised. Additional feedback Aside from several additional comments and suggestions that were raised as part of the data cleanse RFI (summarised in section 2.10), there was a consistent call for MOSL to facilitate further discussion on key areas to better understand challenges and find ways to resolve issues. 1.3 Next steps The responses to the RFI have been used to develop a series of activities as part of a programme of data cleanse work. We will publish a plan setting out these activities at the beginning of March 2021. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 5
Broadly, this plan will include details on the following activities that will be implemented over the next 12 months: Incentivising the completeness of core data items and resolving identified data quality issues through performance monitoring and rectification Developing a process for sharing market data to enable end-users to provide better quality or missing data to the market Investigating and assessing options for incorporating user-verification for core data items to capture accuracy Reviewing the use of meter manufacturer in CMOS validation for meter read submissions Refining the purpose and usage of ‘Customer Name’ and ‘Customer Banner Name’ data items in CMOS. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 6
2. Summary of RFI Responses We would like to thank the trading parties that took the time to respond to the request for information (RFI). The responses consisted of 14 wholesalers and 10 retailers, providing good overall coverage of the market (see Table 1 for further details). Table 1: Overview of respondents to the data cleanse RFI Group Number of respondents Wholesalers 14 Retailers 10 Total Respondents 24 Aggregate supply point share 97 per cent Wholesalers Retailers Affinity Water Limited Business Stream Limited Anglian Water Services Limited Castle Water Limited Bristol Water Plc Clear Business Water Limited Dwr Cymru Everflow Limited Northumbrian Water Limited Pennon Water Services Limited Portsmouth Water Limited Smarta Water Limited Severn Trent Water Limited Veolia Water UK Limited South East Water Limited Water Plus Limited South West Water Limited Water2Business Limited Southern Water Services Limited Wave Limited Thames Water Utilities Limited United Utilities Water Limited Wessex Water Services Limited Yorkshire Water 2.1 Impacts of poor data quality To ensure that we have captured the main impacts of data quality issues in the market, we asked trading parties if they experienced any additional impacts of poor data quality aside from those already identified. Customer and premises data In the RFI, we highlighted the following impacts of data quality issues for customer and premises data: Missing property reference data hinders the verification of premises address data, occupancy status, gap site eligibility and supply point eligibility Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 7
Unreliable customer data obstructs retailers from reliably locating customer supply points within CMOS, creating costs and barriers for tendering and switching customers, as well as reputational damage. Figure 2 shows a summary of the main impact areas identified by respondents for customer and premises data. Of the retailers that responded, 80 per cent (eight) said they experience additional adverse impacts from poor- quality customer and premises data. The additional impacts highlighted by retailers include the following: Increased resourcing and administration cost due to the need to verify the accuracy of premises data or obtain missing premises data; and a high volume of site verification and bilateral requests to wholesalers to verify supply information when this information is unclear Reduced ability to identify, tender and switch customers due to the lack of uniformity and general quality of customer and premises data, particularly for national or multi-site customers Impacted business growth due to the inconsistency of customer data impeding retailers from effectively competing for tenders against incumbent suppliers Poor customer experience and reputational damage due to incomplete or inaccurate supply point data leading to supply points being switched in error; and inefficient verification of occupancy due to difficulties accurately identifying relevant premises Hindered debt recovery due to misaligned CMOS and billing systems data, delaying legal proceedings to verify customer data when there are differences. Most retailers indicated a concern with an impact on customer experience and reputation due to poor data quality. Larger, incumbent retailers were particularly concerned by the added cost of obtaining missing premises data (i.e. UPRN and VOA); and the need to design additional processes and workflows to manage the inconsistent completeness and accuracy of key data items when verifying premises details or occupancy. Smaller retailers were more likely to highlight the impact on business growth and the ability to tender or switch customers. Of the wholesalers that responded, 71 per cent (10) said they experience additional adverse impacts of poor- quality customer and premises data. The additional impacts highlighted by wholesalers included the following: Difficulty identifying and contacting customers for Guaranteed Service Standards payments or following an unplanned event, emergency response or supply interruption – the incompleteness of the Security and Emergency Measures Direction V (SEMDV) flag was noted, suggesting that it should be mandatory Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 8
Water efficiency and market insight is impeded due to missing or inaccurate customer segmentation details, such as Standard Industrial Classification (SIC) codes. Several wholesalers said they are starting projects to improve the completeness of UPRN and VOA data in CMOS; however, some wholesalers questioned the value and expense of maintaining these data items. Our analysis and the responses to the RFI provide strong evidence that UPRN and VOA data are key for fulfilling core market obligations and activities. This includes the following benefits of UPRN and VOA data: It provides a unique reference that identifies a property throughout its lifecycle, from planning permission through to demolition It enhances the ability of trading parties to efficiently verify premises and customer address details and to identify gap sites, demolished sites, occupancy, duplicate premises and erroneous supply points It creates a reliable link with external datasets that facilitates the verification of broader data accuracy and better insight across the market. Figure 2: Summary of main impact areas of poor-quality customer and premises data Market Insight Debt Recovery Customer Service Occupancy Data Cost to Serve Business Growth Premises Administration Customer Visibility 0 1 2 3 4 5 6 7 8 9 10 Wholesalers Retailers Meter location data In the RFI, we highlighted the following impacts of data quality issues for meter location data: Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 9
Difficulty finding and reading meters leads to increased costs for trading parties and could damage reputation Missing reads reduces the accuracy of settlement, leakage calculations and customer billing. Figure 3 shows a summary of the main impact areas identified by respondents for meter location data. Of the retailers that responded, 50 per cent (five) said they experience additional adverse impacts of poor- quality meter location data. The additional impacts highlighted by retailers include the following: Increased cost-to-serve due to meter reading companies struggling to find and read meters based on the CMOS location data Additional Market Performance Standard (MPS) charges due to issues related to data items retailers have no control over Estimated reads are driving non-payment of bills due to perceived over-billing where meters cannot be read – this is leading to lost revenue as customers sometimes refuse to pay or dispute charges based on estimated reads Unable to identify leakage and missed water efficiency opportunities due to inaccurate consumption data. Of the wholesalers that responded, 57 per cent (eight) said they experience additional adverse impacts of poor-quality meter location data. The additional impacts highlighted by wholesalers include the following: Customer disputes and market friction due to missed meter reads contributing to inaccurate estimates, leading to inaccurate end customer billing and inaccurate settlement Increased cost to serve through bilateral forms, alongside additional costs for abortive visits and meter verification Additional costs and poor customer service due to estimation hindering the ability to identify leaks. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 10
Figure 3: Summary of main impact areas of poor-quality meter location data Consumption Accuracy Debt Recovery Cost to Serve MPS Penalties Meter Read Performance Water Efficiency Settlement or Billing Accuracy 0 1 2 3 4 5 6 7 Wholesalers Retailers Meter details data In the RFI, we highlighted the following impacts of data quality issues for meter details data: Increased meter read rejections reduces the accuracy of settlement, leakage calculations and customer billing Inaccurate meter details can lead to increased unplanned settlement runs and query requests, increasing trading party costs Incorrect meter size details can cause the wrong read frequency to be applied, leading to wasted effort and increased costs. Figure 4 shows a summary of the main impact areas identified by respondents for meter details data. Of the retailers that responded, 40 per cent (four) said they experience additional adverse impacts of poor- quality meter details data. The additional impacts highlighted by retailers include the following: Additional costs due to poor meter manufacturer and inaccurate meter dial data through either meter re-reads, additional IT solutions or settlement reruns Inaccurate billing due to incorrect meter dials data, resulting in customer disputes and under-recovery of charges. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 11
Of the wholesalers that responded 57 per cent (eight) said they experience additional adverse impacts of poor- quality meter details data. The additional impacts highlighted by wholesalers include the following: Increased unplanned settlement runs, increased cost to serve and inaccurate customer billing are driven by inaccurate rollover flags Verification of meter manufacturer details can be confusing for third parties to verify due to the lack of uniformity of the data and this can hinder meter reading Figure 4: Summary of main impact areas of poor-quality meter location data Meter Verifcation Read Rejections Customer Service Revenue Cost to Serve Inaccurate Billing or Settlement 0 1 2 3 4 5 6 7 8 Wholesalers Retailers The responses have provided strong evidence for the impacts of poor data quality. They have both confirmed the impacts we had already identified through our analysis and insight gained through subject matter experts; and provided additional insight into the challenges of data quality. In particular, the detail provided by retailers regarding the impacts of poor-quality premises data will help us to prioritise remediation activity as part of the data cleanse work. 2.2 Trading party activities to mitigate poor data quality We asked trading parties what activities they undertake to mitigate the impacts of poor data quality and to estimate the cost of those activities. This information would enable us to build a case for data cleanse work and prioritise the required activity more effectively. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 12
Based on analysis of the responses, we were able to determine an assumed annual salary cost and associated oncosts for a full-time employee (FTE) assigned to data management activity. From the responses from trading parties regarding the number of FTEs addressing data, we were also able to determine a range of FTE cost. We then derived an average cost per supply point of managing data and estimated the total minimum annual cost to trading parties of data quality issues to be approximately £10m per year. Retailers account for approximately 56 per cent (£5.6m) of this cost and wholesalers account for the remaining 44 per cent (£4.4m). Please note that this is a strictly minimum cost as it only considers the resource cost of managing poor quality data and does not include the cost of third-party services to assist with data management nor revenue loss from reputation or business growth impacts. Table 2 provides a summarised breakdown of the costs of specific data quality management activities based on the responses to the RFI. Where possible we have indicated which of the three core data groups are impacted by these activity costs. Table 2: Summary of data quality management costs Estimated Annual Market Percentage Affected Core Data Data Management Activity Cost (£ m) Share Group Locating and verifying meters 3.13 31 per cent Meter location data Verifying premises details and Customer and 2.81 23 per cent onboarding customers premises data Managing system mismatches 1.88 19 per cent All Administrating bilateral requests 1.35 14 per cent All Meter details validation 0.83 8 per cent Meter details Customer and Managing vacancy 0.52 5 per cent premises data Total 10.00 100 per cent - By aggregating these costs, we found that data management activities to resolve data quality issues with customer and premises data accounted for approximately 39 per cent (£3.89m) of the total cost to the market, meter location data approximately 42 per cent (£4.2m) and meter details approximately 19 per cent (£1.91m). Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 13
Customer and premises data Of the retailers that responded 80 per cent (eight) said that mitigation activities requiring additional dedicated FTE resource were required to manage data quality issues associated with customer and premises data. The data mitigation activities for customer and premises data highlighted by retailers include the following: Missing UPRN and VOA data is being obtained and maintained at a cost outside of CMOS as retailers are not the data owner The resolution of poor data provided at supply point registration is adding to onboarding costs: for example, occupancy tracing is an expensive and time-consuming exercise which is being hampered by poor premises and address data Dedicated FTE resource and third-party services are required to manage high volumes of bilateral verification requests and forms due to poor premises data. Of the wholesalers that responded 100 per cent (14) said that mitigation activities requiring additional dedicated FTE resource were required to manage data quality issues associated with customer and premises data. The data mitigation activities for customer and premises data highlighted by wholesalers include the following: Management of differences between wholesaler systems and CMOS FTE and third-party services are being used to investigate and validate occupancy Verification and investigation of premises to maintain UPRN and VOA data Resolution of inaccurate or incomplete supply point data at registration. Figure 5 summarises the main resource areas for mitigating poor customer and premises data. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 14
Figure 5: Summary of customer and premises data management resourcing Premises Registration Bilateral Forms Premises Verification Premises Occupation Systems Data Misalignment 0 2 4 6 8 10 12 Wholesalers Retailers Meter location data Of the retailers that responded 50 per cent (five) said that mitigation activities requiring additional dedicated FTE resource were required to manage data quality issues associated with meter location data. The data mitigation activities for meter location data highlighted by retailers include the following: Resource is being used to address missed meter reads, including the cost and time expended in initial attempts to read a meter and managing additional bilateral requests to resolve issues Skip code management for failed meter reads requires dedicated resource and heavy administration due to high volumes Cost is being expended on market performance charges due to poor meter location data. Figure 6 summarises the main areas of resourcing required for data management activities. Of the wholesalers that responded 100 per cent (14) said that mitigation activities requiring additional dedicated FTE resource were required to manage data quality issues associated with meter location data. The data mitigation activities for customer and premises data highlighted by wholesalers include the following: Management of meter location data differences with CMOS Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 15
FTE resource and third-party services are being used to address long unread meters, including projects to rectify issues with free text fields and Geographic Information System (GIS) coordinates Resource is being expended to manage bilateral requests Sharing of additional data to retailers that currently sits outside of CMOS such Automated Meter Reads (AMR) and additional location data. Figure 6: Summary of main areas of wholesaler meter location data management resourcing MPS Penalties Meter Verification Meter Data Share Bilateral Requests GIS Data Issues Locate Meters Systems Data Misalignment 0 2 4 6 8 10 12 Wholesalers Retailers Meter details data Of the retailers that responded, 50 per cent (five) said that mitigation activities requiring additional dedicated FTE resource were required to manage data quality issues associated with meter details data. The data mitigation activities for meter details data highlighted by retailers include the following: Management of mismatched meter data due to legacy data concerns that accurate asset data was not consistently uploaded at market opening Administration of high volumes of meter read rejections due to inconsistent approaches used by wholesalers to maintain meter details data. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 16
Figure 7 summarises the main areas of resourcing required for data management activities. Of the wholesalers that responded 100 per cent (14) said that mitigation activities requiring additional dedicated FTE resource were required to manage data quality issues associated with meter details data. The data mitigation activities for meter details data highlighted by wholesalers include the following: Resource is being assigned to audit and monitor mismatched meter details, particularly for meter manufacturer and physical meter size – this activity tends to be driven by either bilateral requests or following a meter exchange FTE and third-party services are managing meter verification requests. Figure 7: Summary of main areas of meter details data management resourcing Read Rejections Bilateral Requests Meter Validation Systems Data Misalignment 0 1 2 3 4 5 6 7 8 9 10 Wholesalers Retailers The responses have provided a clear basis for developing a case for data cleanse and will help us to effectively prioritise data cleanse activity. Specifically, the responses have shown that improving the quality of meter location data is a priority, followed closely by premises data. 2.3 Benefits of data quality We asked trading parties to detail the benefits to their organisation of good data quality to gain insight into the desired outcomes of data cleanse work. Figure 8 summarises the main benefits highlighted by trading parties. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 17
Figure 8: Summary of main benefits of good quality data Improved Occupancy Rates Enhanced Market Relations Reduced Customer Complaints Customer Engagement Enhanced Market Insight Water Efficiency Improved Market Reputation Increased Revenues & Profit Consumption, Billing & Settlement Accuracy Customer Satisfaction Improved Efficiency 0 1 2 3 4 5 6 7 8 9 Wholesalers Retailers Both retailers and wholesalers consistently said an improved customer experience was the main potential benefit of improved data quality. They pointed to being able to focus on providing excellent service, significantly improved operational efficiency, more efficient and effective meter reading, a reduction in missed or skipped meter reads and a reduction in read rejections. Trading parties suggested this would lead to more customer satisfaction and trust, better customer experience and shorter customer journey times, lower performance penalties and improved accuracy of settlement, leakage calculations and customer billing. This in turn would generate greater confidence in the market through reliable and stable billing, leading to less customer complaints and improving timely billing and payment. Retailers said that customers would also be more efficiently and easily switched and on-boarded, which would improve retailer reputation and increase engagement with the market. Retailers also said that the benefit from improved operational efficiency, an enhanced ability to resolve bad data and remove unnecessary expense would improve margins and profitability. This would improve competition and increase the scope for providing enhanced customer services, including water efficiency support and advice. Wholesalers said that improved data quality would provide a platform for retailers to provide better services to customers, support vacancy programmes, identify leakages, improve accuracy on primary settlement and less corrective settlement runs. Some respondents noted the opportunity to enhance the relationship with retailers and reduce the number of bilateral requests due to poor data. Wholesalers specifically highlighted the importance of being able to communicate effectively with customers regarding planned work, incidents, disconnections/reconnections, or targeted initiatives, which is greatly enhanced by having accurate customer data. Similarly, better quality data allows a more intelligent emergency response based on insight into the customer and customer types. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 18
Both retailers and wholesalers pointed to reducing the time and effort required to manage meter data queries and to minimise the number of bilateral requests by having better data quality. Wholesalers noted better data quality would also support the provision of timely and accurate responses to retailers, improving the bilateral experience for retailers. The responses have shown that improving customer outcomes is a key objective of data cleanse, both in terms of customer billings but also customer satisfaction and the reputation of the market. Improved accuracy of settlement and lower cost-to-serve has also shown to be important. 2.4 Paid for data services We asked trading parties whether any paid for data services are used to manage or enrich data and how these services were used, their limitations and advantages or disadvantages compared to free-to-use datasets. This would be used to further understand the activities, costs and challenges required to manage data quality issues. A summary of the main responses can be found in Figure 9. Of the retailers that responded, 60 per cent (six) retailers said they use third-party services to manage data quality issues. The types of services include desk-based research, site visits, investigations, and data enrichment. The uses of third-party services included confirming customer or site details (i.e. company name or occupancy status), verification of occupancy and issues associated with the meter and to identify potential occupiers of a property or meters that meter readers cannot locate. One retailer said they are using external resource for analysis of transactions and data to monitor and address issues at a large-scale covering vacancy change applications, backdated amendments and meter read submissions following meter exchanges. Verifying change of tenancy or occupancy was a common use of paid-for-services, and respondents said this is because they typically have better quality information; however, it was highlighted that this represents a significant cost compared to ‘free-to-use’ external data sets, even though it ensures the accuracy and completeness of data. A different retailer clarified that they consider this data cleanse for data items they do not own or control, rather than data enrichment. They said data enrichment is only pursued once the data has been cleansed at a cost. They highlighted, for example, that enhanced occupancy tracing services relies on having accurate address data in CMOS. Several retailers said they are using credit reference agencies (CRAs) to verify the legal entity receiving services. One respondent again noted that the unreliability of the market data set impedes the matching process due to the challenge of establishing the correct customer name or address, particularly for small or medium sized businesses and sole traders. Of the wholesalers that responded, 64 per cent (nine) said they use third-party services to rectify and manage data. The focus was on obtaining more accurate address details, credit searches, verifying supply point eligibility and meter details, obtaining premises identifiers (i.e. UPRN and VOA) at supply point registration, assisting gap site and vacancy initiates and verifying occupancy. One wholesaler noted that third-party services do not Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 19
recognise the water network and the complex nature of water supply, which can significantly reduce the value of enrichment services. Figure 9: Summary of main uses of third-party services used to manage data quality issues Meter Verification Systems Data Misalignment Consutancy Eligibility Verification Credit Checking Finding Meters Premises Verification Occupancy tracing 0 1 2 3 4 5 6 7 Wholesalers Retailers Respondents have highlighted a number of uses of paid-for-services and provided further insight into the costs and challenges of managing data quality issues. This will help to inform our approach to data cleanse going forward. 2.5 Maintaining data quality We asked trading parties how they maintain good quality data in CMOS for which they are the data owner as we wanted to understand the protections in place for maintaining data and whether there were any gaps or inefficiencies. A summary of main responses can be seen in Figure 10. The responses showed a combination of manual and automated processes were being used across the market with varying degrees of proactiveness. Generally, all trading parties were proactively ensuring consistency between their internal systems and CMOS but varied regarding how proactive or reactive there were to resolve known data issues. Most retailers indicated that they predominantly use manual processes to verify and correct data but are moving to automated processes where possible. Most retailers are using consistency reports to identify mismatches between their billing system and CMOS, covering occupancy, customer name and YVEs. Differences are investigated and resolved or updated automatically using the High Volume Interface (HVI) as appropriate. Other retailers have data cleanse teams working to ensure that customer information is accurately captured in CMOS as quickly as possible. These teams also work to correct legacy data. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 20
Many retailers have systems in place that automatically update CMOS following customer notifications, customer switches or a new supply point registration. This ensures that customer details are captured accurately when on-boarded and submitted automatically to the market. Some retailers said they would use site visits to verify occupancy when a customer is unresponsive. One retailer said they use exception reviews and audits to update meter location free descriptors. Reviews can be triggered by skip codes indicating that a meter cannot be found or when a customer or wholesaler notifies of a change in meter location. Another retailer said they rely on their meter reading service provider to notify them of metering data issues. They will then either correct the data themselves or notify the relevant wholesaler. They note that this is time consuming, expensive and suggest that it may not have been a priority for retailers. A different retailer also said they find it “frustrating” to rely on wholesalers to update certain items, particularly meter location data such as GIS coordinates. Most wholesalers said they proactively compare CMOS data to internal systems to ensure consistency using automated processes. In addition, there are reactive procedures to resolve issues or anomalies identified through quality checks, site visits, customer complaints, bilateral requests, dashboards, or periodic reviews. Dedicated teams are in place in most cases, as well as automated checks using external datasets. Some wholesalers said they use technological solutions to capture GPS data when reading or exchanging a meter, which is then provided to retailers and the market. Another wholesaler said that they can use the serial number of newer meters to verify meter details such as number of digits and this is pre-populated on systems. Figure 10: Summary of main activities to maintain good data quality Reactive management Exception Reporting 0 2 4 6 8 10 12 Wholesalers Retailers The responses have shown strengths in several areas of data quality maintenance, especially the aspiration for proactive improvements to data quality and automated consistency checks between internal systems and CMOS. The responses also highlighted areas for improvement, such as: a high level of manual processes in Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 21
place for some retailers, and a reliance on meter reading service providers for correcting meter data that is often expensive and time consuming. 2.6 Data outside of CMOS We asked trading parties if there are any known differences between the core data stored on their own systems compared to CMOS; and, if so, to provide details and indicate whether they would be willing to share this data with the market. This information will help us determine whether there is a need to provide additional routes for data into the market dataset to improve completeness and accuracy. See Figure 11 for a summary of the main reasons for differences between systems. Of the retailers that responded, 70 per cent (seven) said there are differences between their systems and CMOS. Retailers pointed to misalignments in data where they are the owner due to transaction failures, manual intervention, backlogs, system migrations and unintentional data enrichment. Many retailers indicated they are working to correct these issues and reviewing procedures, notifications, and technical solutions to ensure greater consistency. Two retailers explicitly indicated a willingness to share misaligned data for which they are not the owner, such as GIS coordinates and meter location notes; whilst one retailer explicitly said they would have no interest in sharing misaligned data as it could impact their profitability. Of the wholesalers that responded, 57 per cent (eight) said there are differences between their systems and CMOS. Most wholesalers said that there are minimal differences between core data items and their own systems due to the automated processes in place verifying consistency. Many, however, said they do hold additional data that could be valuable to the market. This included photos of meter locations, meter location notes, logger identification and ownership detail and more granular meter read data (e.g. monthly fixed reads), non-billable premises and premises categorisation information. Most wholesalers commented that they share this additional data with retailers where possible, but many added that it would be beneficial to retain this data directly in CMOS. One wholesaler said that in certain instances they have customer details that are not shown in CMOS, which they provide to the retailer subject to appropriate GDPR and Data Protection Impact Statements (DPIA) approval. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 22
Figure 11: Summary of main reasons for system data differences Customer Name Differences Occupancy Status Differences Additional Data Manual Processing 0 1 2 3 4 5 6 7 Wholesalers Retailers The responses provide evidence that data-users do not consistently have proper routes to share data into the market where they are not the data owner. Substantial improvements to data quality are possible if processes can be agreed to allow this data to be uploaded, particularly for key areas such as meter location data. 2.7 New technology and end-user verification We asked trading parties whether they agreed with MOSL implementing new technology to supplement the market dataset with end-user input. This could include input from meter reading service providers to verify the accuracy of core market data items such as meter location and meter details data. Overall, 58 per cent (14) of the respondents said they supported the use of new technology to capture end- user input and 42 per cent (10) were not. Of the retailers that responded, 60 per cent (six) were supportive. The remaining 40 per cent (four) retailers that were not supportive provided additional feedback showing a mixture of viewpoints. One retailer that disagreed said that responsibility for data quality resides with the data owners, and MOSL should instead focus on incentivising and rewarding data quality performance. Two retailers that were not supportive said they broadly agree in principle with capturing end-user input and could be supportive if they had more information. In particular, more detail on the potential technical challenges and limitations would need to be considered before they would support this type of approach. They welcomed further discussion or development in this area. Another retailer argued that retailer-owned data was a competitive element of the market and supplementing the market dataset with this data would reduce competition. They also stated that the purpose of the market Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 23
dataset and CMOS is to provide efficient market operation, calculating settlement and enabling switching. They also noted a concern that maintaining this data could increase the cost of maintaining CMOS, thereby reducing profitability of retailers. Of the wholesalers that responded, 57 per cent (eight) were supportive. The remaining 43 per cent (six) wholesalers that were not supportive provided additional comments. Four wholesalers that disagreed said they would need to see more details but were interested in the concept and supported further exploration, potentially through a working group. It was highlighted that there would need to be appropriate governance to maintain data ownership. It was also suggested that it would need to be mandatory if it was implemented to ensure it added value. Another wholesaler raised concerns about MOSL funding new technology given that trading parties are obligated to ensure accuracy of data. They suggested that the approach for funding new technology or processes should be based on the “polluter pays principle”. One wholesaler said this approach created a risk of data quality deteriorating further as it would weaken accountabilities for ensuring accurate data. They argued that existing market processes were sufficient to update or validate inaccurate data. The responses have provided a strong mandate to investigate the possibilities of incorporating end-user input to capture the accuracy of data. This would enable accuracy to be measured and incentivised, however the challenges need to be explored and discussed with the market. 2.8 Customer details definition During our analysis, we uncovered potential ambiguities in the interpretation of customer data fields. Therefore, we asked trading parties to explain how they are interpreting the customer name data fields, i.e. ‘Customer Name’ (D2027) and ‘Customer Banner Name’ (D2050) as defined in CSD 0104 (‘Maintain SPID Data’, section 4.1.1) to understand whether clarification was required. The majority of trading parties said they interpreted Customer Name as the legal entity of the customer that is responsible for paying for the water consumption or services provided by the retailer. Many retailers said they apply this, as appropriate, to an individual, company or organisation, whereas some retailers only specified a company. Some retailers specified they are using the name contained in Companies House data, but they did not say how they treat customers that sit outside of the Companies House dataset, such as sole traders. There was less consistency for the interpretation of the Customer Banner Name field. Some retailers said they provided the name of the customer’s business or trading name if it was different to the Customer Name (i.e. legal entity); whereas another retailer pointed to the need for flexibility as retailers sometimes need to enter a specific name requested by the customer. They suggested an additional field would be required to capture the customer requested name, but they questioned whether this would be cost effective to maintain. Several respondents suggested the interpretation of these data fields has not been consistent and requested further clarity on their application. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 24
Respondents identified the following usages of the customer names fields: It identifies the legal entity with liability of charges, payment of bills or maintaining contracts It supports meter reading and customer or premises verification by enabling the correct premises to be identified It provides a basis for linking customers that are part of a group with a parent company. The responses have shown that the ambiguities for the customer name data fields, particularly Customer Banner Name, require clarification to ensure they provide value to the market and avoid wasted effort. 2.9 Meter manufacturer validation We asked trading parties whether they agreed with removing the CMOS validation requirement to provide a meter manufacturer when submitting a meter read. This is due to growing evidence that it is causing legitimate reads from being submitted into the market and we wanted trading party views on what should be done to resolve this issue. The majority of respondents were supportive, with 67 per cent (16) of trading parties agreeing with the proposal. Of retailers, 70 per cent (seven) were supportive; and of wholesalers, 64 per cent (nine) were supportive. Trading parties that were supportive argued that the meter manufacturer validation rule provides no benefit and hinders the submission of legitimate reads into the market due to misspellings and case-sensitivity. The wholesalers that were not supportive said they did not believe meters could be uniquely identified consistently without the meter manufacturer validation requirement; or else they would want assurance that meters could be uniquely identified after removing the validation requirement. One wholesaler requested that MOSL provide an estimate of the volume of meter read submissions that fail due to issues with meter detail; and confirm whether this is a market-wide problem or specific to certain trading parties. Several wholesalers argued that the manufacturer names should be standardised instead, potentially using a drop-down option. One wholesaler noted that inclusion of meter manufacturer does not always protect against duplication due to the lack of checks on erroneous removals of meters that are then added in back into the market. They argued that removing the manufacturer validation check would add to this issue and decrease confidence. They suggested standardising the meter manufacturer field and instigate a data cleanse to migrate legacy manufacturer names to the standardised list. They also suggested introducing a unique identifier for all meters to remove the risk of duplication. The retailers that did not support removing this validation check argued that it was key to ensure good quality data is maintained. There was support for reducing the case-sensitivity and improving standardisation. A retailer that supported removing the validation rule also specified that this should not lead to a significant change in the meter submission process or create substantial implementation costs. Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 25
The responses show that a solution is needed to reduce the volume of rejections of meter read submissions and issues associated with inconsistent or inaccurate meter manufacturer data. Further work and consultation are required to achieve consensus, however, as there were a mixture of views on the best way forward. 2.10 Additional feedback Many respondents provided additional, detailed comments or suggestions. All points have been taken into consideration, with some having been directly incorporated into relevant sections above. Those not incorporated have been summarised below: The technology exists to improve or correct GIS coordinates, for example systems such as what3words or Pinpoint Wholesalers should be able to directly update the location code rather than this having to be done by the retailer The inclusion of skip codes would give visibility of trading party attempts to read meters and the reason why they have failed to obtain the read There is no way to identify known accessibility issues with a meter, therefore the volume of problem meters cannot be tracked or easily resolved There could be benefit creating an additional field to indicate if a meter can be read with or without the need to contact the customer Analysis of how current service levels and subsequent market penalties may be impacting data quality could be beneficial Given the work involved in addressing the quality of the data, wholesalers will need to be incentivised with both the incentive of recognition of strong performance and performance rectification measures The impact of COVID-19 should be considered as part of any data cleanse activity and proposed timelines for completion Verification of data needs to be carefully considered against competition as some data items are competitively procured The fairness of MOSL incurring costs for data cleanse should be considered given that some retailers have already incurred costs to resolve data issues and would also contribute to MOSL’s costs The question of whether the quality of data items is part of the competitive element of the market also needs to be considered Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 26
The facility for photos of meter locations and reads to be uploaded to CMOS, or another shared system between wholesalers, retailers and third-party meter readers would be beneficial Reconsidering the eligibility of sites such as troughs, places of worship, public convenience and standpipes that may be problematic for the end-customer after four years since market opening Comparisons of data sets that cannot achieve at least a 95 per cent match should not be used to indicate trading party data quality All data cleanse activity should have a clearly intended purpose and outcome to ensure that resourcing and funding is focused on improving the data items that will really make a difference MOSL should engage with trading parties on a one-to-one basis to provide a clear explanation of the checks which are being applied to data quality measures before any public comparisons are published References to the timeliness of data and looking at the percentage of data that has been maintained or validated within a certain timescale does not take account of checks made to data that is found to be accurate and therefore would not result in a change in the market data The process of updating VOA and UPRN data will take resource away from long unread, vacant, unpaired SPIDs, stopped meters; and therefore, there needs to be a clear case MOSL is encouraged to facilitate further discussion to continue capturing challenges and potential inefficiencies caused by data in the market The time and cost of incorporating third-party data items, such as UPRN in the market dataset, should not be overlooked when mandating requirements to update and maintain this information It is not clear how MOSL propose to ensure that trading parties are updating CMOS with the same data that is held in their systems in a timely manner Verification of data needs to be carefully considered against competition Careful consideration of the cost of using third parties to verify data is required, especially for retailer- owned data as any costs incurred by MOSL on this are paid for by trading parties. If not all trading parties have invested not this is going to be “unfair” A clear cost benefit analysis should be made before any services are procured to determine that the additional benefits to the market outweigh the cost with clear decision points and governance Further clarification of the definition of data items can be of benefit The transaction to update the occupancy status of a premises should require updated customer name details or it should be automatically set to ‘NULL’ or ‘NO CUSTOMER’ Summary of Responses to the Core Market Data Cleanse Request for Information (RFI) Version 1.00 | 17 February 2021 27
You can also read