Managing debt in the Utility industry has inherent challenges not faced by other sectors. For starters the product itself is fundamental to its customers’ needs. Disconnection of supply is neither a quick nor lightly-regarded outcome for a supplier. By which time the account will have moved even further into the red. Suppliers and consumers are also facing a challenging economic environment, with likely interest rate increases on future debt. However, by introducing data science techniques we are able to better inform collections decisions, creating a more effective way for these companies to manage and reduce their debt books.
To recover debt, suppliers use increasingly costly activities across the collections (or dunning) paths, ranging from letters to outbound calls, site visits, legal action and finally a meter exchange (in energy retail). These measures typically involve upfront costs irrespective of success outcome, therefore it’s important that suppliers ensure they are targeting the right steps towards the right customers (ie can’t, versus won’t pay).
There are various tools available to ensure the right interventions are adopted as part of a credit risk strategy and debt recovery approach. As the quantity of data available for each customer increases, suppliers can build a detailed individual profile, and identify which accounts they should be targeting: inputs typically include geographical, personal and socio-economic data such as credit score. Data quality plays an important role. The accuracy and completeness of the information is a key factor in the model and determining credit risk. The more information fields obtained, the more detailed the profile. However, it’s not uncommon for suppliers to face gaps of missing or incorrect customer data when trying to assemble this view.
So how can we feasibly enhance the existing data set?
The answer could lie with thinking (then retrieving) data from quite literally outside of the box. Internally-owned data may not always be sufficient to build a model with satisfactory performance. Publically accessible “open data” sources such as those published by the Office for National Statistics (ONS) can help fill in these gaps. In recent analysis we have made use of the deprivation index, which describes the average resident’s socio-economic status via 33,000 specific areas measured on education, health, or income. This extra lens provided a detailed view of the likely pockets of very low or high collectability across the country. The ONS is one of several open source databases and has an extensive coverage of data, ranging from business and economics to demographics, with applications suitable beyond just the utilities sector.
Suppliers can typically improve the value from their existing data, but there is an ever-increasing level of additional customer insight available through the free public datasets available. Whether the aim is to improve operational activity or simply to obtain better understanding of who their customers are, the data’s there – why aren’t we making the most of it?
We're keen to hear your views on this issue - or issues relating to it.Contact us if you have a view on this that you would like to share.