May 4, 2019
Why Limit the Benefits of Machine Learning to the Front-Office?
This article was originally published in the Spring 2019 issue of What’s Next magazine, published by TMG Consulting.
Machine learning (ML), and AI in general, has been celebrated for front-office and customer service benefits. With the introduction of smart meters, many ML use cases focus on leveraging a large volume of data for analytical purposes, primarily improving network performance and reducing commercial losses. Meanwhile, there has been little focus on the benefits of ML in the back office, though every utility understands that improvements in back-office operations go straight to the bottom line.
Here are Three Use-Cases in Back-Office where ML can Add Significant Value
Customized Billing Validation
In most utilities, customers are grouped by type of customer (residential or commercial) or by customer size (small and large commercial). A fixed set of validations is defined for each of these segments and is used during the validation process. This results in false positives for customers who are on the fringe and for those whose circumstances have temporarily changed due to weather, maintenance downtime, vacation, etc. With ML, we can look at customer-specific and external data to build real-time checks and tolerances that are specific to each customer. This results in utilities being able to prevent incorrect bills from going out, reduce field visits to recheck data, and improve revenue collection because you’re not holding back invoices for validation.
Using ML to recognize patterns and match payments to the correct account has shown significant improvement in accuracy. This becomes especially important in deregulated markets where the number of participants is many and increased efficiencies quickly deliver value.
Preventing Complex Process Errors
Using the power of ML to identify hidden patterns and relations between seemingly disjointed processes as well as predicting errors which have yet to occur and being able to make proactive corrections is another area where ML shows huge potential.
Rest assured, excellent use cases for machine learning don’t stop in the front office.