By Eugene Hamrick, Rappahannock Electric Cooperative (REC) and BrilliT & Tomer Borenstein, BlastPoint.
For many utilities, customer satisfaction has suffered over the past two years. The rise of extreme weather-related outages, the pandemic, economic uncertainty, and higher monthly bills have amplified consumers’ displeasure. J.D. Power reports that overall satisfaction with electric utilities fell 17 points in 2021.
Utilities that prioritize customer satisfaction should consider implementing these 7 AI-enabled solutions in 2023.
1. Outage Management
Multiple surveys indicate that customer satisfaction is heavily influenced by utility outage activity and communications. Under normal conditions (blue sky), customers expect to be notified of an outage within 30 minutes. During stormy weather, this timeline is extended to 4 hours. In any case, these timeframes present a significant challenge for overburdened communications teams. Meeting customer expectations will become even more difficult if the number of major weather events intensifies, as predicted.
Utilities can better prepare for outages using deep learning to create models that predict outage duration under different conditions per area. Using historical grid, crew, weather, AMI, outage, and other relevant data, AI weather models can predict the length of future outages and the best timing of notifications. When integrated with SMS and other communication technologies, AI outage models can help utilities notify customers when service will be restored. With deep learning AI models, utilities can expect to achieve storm outage prediction accuracy as high as 90%.
2. Smart Automation
Large Language Models (LLMs) represent one of the decades’ largest leaps in AI and ML. Entering mainstream awareness in late 2022, LLMs like ChatGPT bring human-like qualities to conversations and text-based applications, such as natural language generation, dialogue generation, natural language understanding, and natural language processing. As a result, LLMs are a powerful tool for customer care and communications teams that are stretched thin.
The application of LLMs is nearly infinite. A few of the most straightforward applications for utilities include data cleanup and processing, generating personalized communications for customers, and improved automated customer service.
3. Customer Satisfaction (CSAT) and Net Promoter Scores (NPS)
Common uses of rolling customer surveys to measure CSAT and NPS scores are limited. Such surveys typically have low temporal granularity (quarterly or monthly), low spatial granularity, and are prone to survey biases. AI-Driven CSAT and NPS scoring overcome these limitations by combining survey data with operational utility data. AI is able to learn patterns and score customers individually, every day.
With automated CSAT scoring, utilities can intervene with dissatisfied customers early. Further knowing which programs impact CSAT and NPS the most allows utilities to prioritize resources and take meaningful action. Using appended survey data, AI models can provide a satisfaction score per customer daily.
4. Customer Segmentation
Consumer segmentation has been used somewhat widely by utilities over the past few years, but many of these off-the-shelf solutions are inaccurate and hard to activate. Alternatively, objective-driven customer segmentation can be tied to a specific goal, like a customer-facing program or initiative. To boost accuracy and achieve better results, AI models dynamically generate micro-personas that are optimized for a particular objective. AI-based propensity modeling and semi-supervised clustering generate the segments, then identify which customers are most likely to take a desired action. This allows utilities to target customers based on data profiles, using the most effective channels and messaging for each segment.
Objective-driven segments help utilities serve specific customer-based objectives. For instance, knowing which customers are most likely to go electric, whether that means buying an electric vehicle, stove or heating system. Segments can also help direct customers away from clogged call centers and toward self-serve resources.
5. Billing & Payment Engagement
Utilities are facing alarming levels of consumer debt due to the lingering pandemic, rising energy costs, inflation, and the uncertainty of a looming recession. Using propensity and segmentation models to drive enrollment in financial assistance programs like LIHEAP, income-eligible energy efficiency, and other billing programs can help utilities keep bills low and payment-in-full rates high.
Propensity-to-pay models allow utilities to identify which customers are most likely to pay in full and increase collection rates, even during moratoriums. For example, a utility used AI models to double the number of customer assistance grants given out, while another achieved 20% enrollment rates for commonly underutilized assistance programs.
6. Program Equity
Many datasets contain cultural biases that create a barrier between consumers and energy equity. AI is useful to identify and disrupt dataset biases, which then makes it easier to reach underserved customers. Using third-party demographic data to append customer data helps utilities understand the various types of energy inequity their customers may be experiencing. With auditable AI models, utilities can take steps to reduce inequity by creating an “equitability score” for specific initiatives.
Take as an example, the energy equity hot topic of electric vehicle charging. Using AI models to predict propensity for EV ownership balanced against economic data helps utilities decide where to place electric vehicle chargers and avoid placing charges exclusively in areas preferred by wealthy customers. Additionally, AI helps utilities plan for charging where there is limited load capacity.
7. Data Access & Literacy
To provide modern customer experiences, utilities must bridge organizational data silos. Customer intelligence (CI) platforms make it easy to make data-driven insights interoperable across the organization. When secure self-serve access to data is granted that data becomes truly actionable. The ability to work with data from multiple systems in one unified environment allows utilities to compound the ROI of data and deliver a consistent customer brand experience across departments. Customizable security levels help utilities keep sensitive customer information private and maintain compliance with applicable regulations.
CI platforms help utilities achieve data maturity and drive a customer-centric culture. When data is removed from IT siloes, utilities gain the rewards of engaged and satisfied customers, lower consumer debt levels, and recognition as market leaders.
Ready to Get Started with AI?
Utilities that are interested in boosting customer satisfaction through modern customer experiences, better engagement strategies, and more efficient operations should look to AI as the answer. AI has already helped many utilities drive measurable financial and operational results and deliver the personalized experiences consumers expect, including time-sensitive and on-target communications.
About the Authors
Eugene Hamrick is Director of Enterprise Analytics and Innovation at Rappahannock Electric Cooperative’s (REC) and BrilliT, where he serves to promote data enhanced decision-making and analytic solutions that achieve maximum benefits from the use of data assets.
With several years of electric utility knowledge and data science experience, Hamrick brings a passion to solving the bigger business problems with actionable data insights and has a unique perspective on the challenges facing electric distribution cooperatives. In 2022, he was designated a Top 25 Thought Leaders in Utility Analytics by Utility Analytics Institute.
Tomer Borenstein is Co-Founder & CTO of BlastPoint. Borenstein is credited with generating millions of dollars in sales and securing partnerships with a growing list of national and international corporate customers. He has been named a Business Times’ “30 under 30,” has authored numerous articles and speaks regularly at conferences on the benefits of AI and machine learning for customer engagement and business growth. His previous work in the startup space includes an exit to Apple. Tomer holds a Bachelor’s degree in Electrical and Computer Engineering and a Master’s in Machine Learning and Computer Vision from Carnegie Mellon University.
