This special interest group is where customer care professionals share tactics on how utilities are improving interactions with their customers. 

WARNING: SIGN-IN

You need to be a member of Energy Central to access some features and content. Please or register to continue.

Post

Machine Learning Reshapes the Customer Care Landscape

Utilities now collect more information about their customers than at any time in history, and machine learning is emerging as a key way to empower them to leverage the data for competitive gain. As a result, machine learning is changing how these organizations respond to customer requests for the better.  

Customer service has been driven by data analysis and nowadays utilities have lots to examine.  In 2020, the world will create 44 zettabytes of information up from 10Z bytes in 2015, according to market research firm International Data Corp. (IDC). To put those numbers into context, one zettabyte of apples would fill the Pacific Ocean.

Collecting data is the first step in a process to leverage information and improve customer care. By itself, data does nothing. Its value only comes once corporations garner insights and use them to improve business processes. For years, energy providers tried to use data to better understand customer needs, and machine learning, the next rung on Artificial Intelligence ladder, has that potential.

Cutting Down on Paperwork

Leveraging information is labor-intensive. Traditionally, customer service managers had to sift through reams of reports, correlate items in an ad-hoc manner, make business decisions, and monitor their results. With data volumes growing, executives have been drowning in information.

Machine learning offloads much of the gathering process. The technology relies on algorithms to illustrate data point connections and generate reports that help utility execs identify trends, initiate changes, and streamline business processes.

Improving Customer Care

Customer service representatives have long thirsted for real time information, so they could respond to customer needs as soon as they arise. Previously, businesses lacked the compute power needed to sift through the countless possibilities that could occur at any moment. Currently, with the advent of massive cloud based data centers, high speed communications lines, and machine learning, that goal becomes possible. In fact, 57% of executives believe the most significant benefit of machine learning will be improving the customer experience, according to Forrester Research Inc.

Here, companies are taking multiple tracts to leverage machine learning to reach that goal. Quick resolutions to service calls is something that both companies and customers desire. Vendors currently collect lot of information about each customer interaction. Rather than simply continue to respond to seemingly random queries, enterprises are building models that anticipate what the consumer is looking for and respond proactively rather than reactively. For instance, if a client picks up the phone after clicking through your Web site, the contact center representative deduces they have been unsuccessful in finding needed information.

Recent technical advances enable companies to collect oodles of customer information. Machine learning is empowering them to use that data to improve customer satisfaction, reduce costs, and increase revenue. As a result, the emerging technology is becoming a key cornerstone in today’s customer service center.

 

Paul Korzeniowski's picture

Thank Paul for the Post!

Energy Central contributors share their experience and insights for the benefit of other Members (like you). Please show them your appreciation by leaving a comment, 'liking' this post, or following this Member.

Discussions

Bob Meinetz's picture
Bob Meinetz on Jan 16, 2020 5:22 am GMT

"Rather than simply continue to respond to seemingly random queries, enterprises are building models that anticipate what the consumer is looking for and respond proactively rather than reactively."

Caller: "Hello, I'd like to know whether Anytown Electric buys its gas to generate electricity at fair market value. I'm asking, because both Anytown Gas, Inc. and Anytown Electric, Inc. are wholly-owned subsidiaries of Anytown International, Inc., and it seems your gas subsidiary might be marking up its price to..."

Artificial Care Representative: "Hello, and thanks for calling! It seems you want to know your bill balance. Is that correct?"

Caller (peeved): "REPRESENTATIVE."

ACR: "OK, you'd like to speak with a Customer Care Representative! I can understand what you say, and even anticipate your needs. So to put you in touch with the representative who could best help you, I need to know a little more information about why you're calling. You can say, 'Bill Balance,'or 'Question About My Bill', or 'Clean Energy Plans,' or 'Free LED Bulbs,' or..."

Caller: "TECHNICAL SUP...

ACR (interrupting): "OK, you'd like to speak with a Customer Care Representative! I can understand what you say...

Caller: "TECHNICAL SUPPORT."

ACR, after pause: "OK, you have a Seemingly Random Query. All of our Seemingly Random Representatives are busy right now, but your query will be responded to in Seemingly Random Order!"
(Vapid music on line, caller hangs up).

ACR, to no one in particular: "If you'd like to take a Brief Customer Survey..."

"Improving Customer Care

Here, companies are taking multiple tracts to leverage machine learning to reach that goal. Quick resolutions to service calls is something that both companies and customers desire..."

Paul, the goal of machine learning is not improving customer care but saving money. Instead of hiring and training customer service representatives, investor-owned utilities are thrusting the burden of convenience onto customers. I'll go out on a limb and suggest not one utility customer would say automated customer care resolves service calls more quickly than a trained representative.

The difference between a bank or a utility monopoly making that choice is, of course, customers can switch banks. A more productive tract to satisfying utility customers: a system that tracks the number of minutes wasted waiting for reps to respond to a customer's request, and deducts some appropriate value (50¢/minute?) from the customer's bill. Any progress on that front?

Matt Chester's picture
Matt Chester on Jan 15, 2020 10:24 pm GMT

Paul, the goal of machine learning is not improving customer care but saving money. Instead of hiring and training customer service representatives, investor-owned utilities are thrusting the burden of convenience onto customers. I'll go out on a limb and suggest not one utility customer would say automated customer care resolves service calls more quickly than a trained representative.

Improving customer care does save money-- the more easily customers can find answers themselves, the less likely they'll be to call into a service center, which will inherently reduce customer service costs to the utilities. Does that transfer burden to the customers? I suppose. But I also suppose if there's a tool like an online energy portal that quickly and easily answers the bulk of questions customers call in about that they'd prefer to have that rather than have to take the time and effort to call in. 

Bob Meinetz's picture
Bob Meinetz on Jan 16, 2020 12:42 am GMT

"But I also suppose if there's a tool like an online energy portal that quickly and easily answers the bulk of questions customers call in about that they'd prefer to have that rather than have to take the time and effort to call in."

Could be a generational difference. Twenty years ago getting a competent customer service rep on the line was easy, and much faster. But it did require interacting with another live human, speaking in complete sentences, etc. I guess that could be considered effort.

Paul Korzeniowski's picture
Paul Korzeniowski on Jan 16, 2020 12:28 pm GMT

These advances have multiple goals. Of course, companies want to reduce operating costs, and having machines take on work that individuals do has been occurring since the first computers were built more than 50 years ago. Then, a roomful of individuals would load punch cards into machines.

The challenge with AI and ML solutions is developing algorithms that accurately understand what the user inputs and respond appropriately. They can handle straightforward requests, like providing them with their balance or scheduling a turn on/turn off date, but struggle with other individual interactions.

The willingness to interaction with these new systems varies dramatically by age. Millenials prefer a text exchange to a phone call. Boomers like to talk to a person.

 

Bob Meinetz's picture
Bob Meinetz on Jan 16, 2020 8:18 pm GMT

Agree Paul, and we're on the edge of a much bigger discussion about the value of AI, and whether there's a limit to whether usefulness and even safety inevitably suffer more than benefit by removing human interaction from important functions we perform every day.

I believe they do. But during a holiday discussion with my sister, she disagreed so vehemently I felt my personal safety was in danger (maybe that discussion would have been better handled by email).

Matt Chester's picture
Matt Chester on Jan 16, 2020 12:47 pm GMT

If in the time it takes me to look up the number for and call my utility's customer service center (not to mention the wait time that inevitably comes) I can instead just as easily look up the answer to my question, I'd consider that a win. I'd consider that a win-win for customer and utility! Of course for more complex questions/issues from customers, calling will still be necessary and so the calling infrastructure is important-- but my point is the key is to weed out those easy to look up quick answers.

As a crude, crude comparison, if Google had an option to call instead of search-- would I rather call or search to ask about the capital of Bulgaria? I'd rather go through the step of Googling Bulgaria and getting the answer myself then finding the phone number for Google, calling them, likely waiting for the next available agent, and then having that agent tell me the answer I could have found just as easily myself. Now if instead I was curious about a nuanced aspect of Bulgaria's history or politics, I might prefer to talk to this imaginary Google agent who can interpret and understand exactly what I'm looking for whereas Google searches won't as directly get me to my answer. 

What I see these customer service machine learning tools as doing is finding out how to expediently answer the capital of Bulgaria type questions without using up resources of human agents while more efficiently directing the history of Bulgaria type questions to the right, trained agents. 

Get Published - Build a Following

The Energy Central Power Industry Network is based on one core idea - power industry professionals helping each other and advancing the industry by sharing and learning from each other.

If you have an experience or insight to share or have learned something from a conference or seminar, your peers and colleagues on Energy Central want to hear about it. It's also easy to share a link to an article you've liked or an industry resource that you think would be helpful.

                 Learn more about posting on Energy Central »