EDF has already installed 2 million smart meters and visits over 8,000 homes every week. But some of these visits result in failure which the organisation needs to respond to. Previously, its focus had been on failure prevention rather than failure rectification. Unless it could acknowledge a problem, explain to customers what would happen next and rebook an appointment, it could be penalised for failing to meet its regulatory requirements, potentially leading to a loss of licence.
To address this, EDF created a single point of reference for all of its failure management activities. This failure management data repository (FMDR) was a custom-built data lake hosting unstructured data from multiple sources. Using AWS, a data mart, data lake and data querying platform was built in which data was cleansed, transformed and unified. This allowed for rapid deployment and scaling, integration into operational environments, governance and exploratory analysis.
Failure data in FMDR was categorised by root cause and now supports outbound communications, including telephone appointment setting, email and direct mail explaining next steps to customers experiencing a failure. Over 120,00 customers have been targeted this way, with one key benefit being the avoidance of customer-driven failures – this has the potential to save over £1 million in operating costs.
FMDR also provides fully-auditable evidence to help satisfy regulatory requirements. Forecasting models for propensity to fail have been built which are shaping yearly targets and installation numbers which have to be accounted for to OFGEM and BEIS. This has led to the government department describing EDF’s approach as “industry leading” and “on another level compared to what anyone else is doing”.
