A century ago, the thought of customer support receiving a helping hand from bots and machines was science fiction. The reality, however, is that humans are just as critical as AI to the future of support. As AI and machine learning models mature, how can organizations use the opportunity to synthesize agents alongside it to deepen insights into customer behavior?
Artificial intelligence is quickly transforming the way businesses operate and people interact with the world around them. It’s also fundamentally changing how companies approach customer support.
According to Statistica, the call center AI market will grow from around $800M in 2019 to around $2.8B by 2024. Gartner predicts that 25 percent of all customer service operations will use virtual assistants by 2020, and that, in the coming year, organizations will invest more in AI-driven bots and Chatbots than they will in mobile apps.
So, do these forecasts mark the demise of the human customer service agent? Is the future of customer support 100 percent driven by machines?
Not even close.
Gartner predicts that AI-related jobs will see steady growth starting in 2020. Instead of replacing humans, AI is expected to augment existing jobs and improve productivity—enabling people to do their jobs more efficiently.
Even AI scholars agree that, as advanced as AI has become, it can’t do everything that humans can do.
“I think, sometimes, there’s a perception that the machines are going to take over and replace all of us, when what organizations actually need is a machine-human symbiosis,” said Professor Raymond J. Mooney of the Department of Computer Science at the University of Texas at Austin. “AI is better than people at certain things, but clearly worse at other things. We need to find the right blend of AI, automation and people to get the best outcomes.”
Support operations that can strategically apply AI and machine learning to empower their agents to have an opportunity to significantly improve the customer experience—from less wait times, faster resolution, to a more personalized level of care. But only if they recognize that as impactful as AI can be, it is not a magic bullet. It’s technology reliant on human beings to reach its full potential.
Strategically Applying AI: Begin with the Goal and Risk
Although AI and machine learning have a number of potential applications in the customer support space, there’s not one “right” answer on where and how to use the technology. It depends on a number of factors, including what the organization wants to accomplish, as well as the level of risk associated with using automation toward that goal.
For example, let’s say that an organization wants to improve its customer experience and overall satisfaction ratings. The biggest complaints are around speed—the speed at which the customers’ call, chat or text is answered, and the speed at which the agent is able to solve their problems.
So, the organization first considers using AI for chat, but determines that carries too much of a risk if the bot gets the answer wrong.
Instead of simply automating a channel, a better approach might involve using AI to provide agents with coached decision-making as they’re having a conversation or a chat with the customer. Instead of the human agent combing through the knowledge base to extract the data, the goal is to use AI to recognize what the customer is asking for, retrieve the appropriate information, and come up with a potential solution for the agent.
All of the data gathering happens invisibly to the customer. The agent reviews the information, with the power to override it if the machine comes up with the wrong solution. The more feedback AI gets, the more it learns, until it becomes proficient enough to become a standalone channel option—still supervised by a real human being.
“When thinking about the application of artificial intelligence and the relationship between human agents and AI, I think a lot of people get that relationship backwards,” explained Greg Melia, CAE, CEO of Customer Experience Professionals Association (CXPA). “It’s not ‘How should AI replace the role of the human agent?’ It’s how can AI help customers get the answers they need at the right time? How can having common and ordinary questions automatically answered free up the human agents to focus on higher-level challenges and situations that require more thoughtful action?”
With every potential change that AI and machine learning brings to customer support, it’s critical to remember two things:
- The best AI applications elevate the human agent, not replace the human agent.
- Every decision on how to deploy AI has to be customer-centric. The question should never be “Will this reduce costs?” but, “Will this change benefit the customer?”
Personalizing the Customer Support Experience
One of the most common uses of AI and machine learning in customer support centers today is automating the initial response, so the call, text or chat is answered more quickly. The virtual agent gathers the basic information and quickly passes that customer, and the gathered data, to a human agent.
But, AI as a gatekeeper is just the baseline.
With today’s technology, organizations also have the opportunity to personalize the entire customer experience to improve satisfaction, loyalty and elevate the brand.
This process starts by using acquired customer data to build customer profiles, and matching individuals with like characteristics and preferences to others who behave in the same way. Using this client data, their recent interactions, sales history, product history and behavior, AI can recognize patterns and make predictions on which channel, as well as which available human agent, is the best match to service each customer.
For example, if Customer X seems to prefer less human interaction, and “looks” (profile-wise) like other customers who are successful with chat, when she logs on the support page, the chat button is larger and highlighted.
If the organization has an automated option, a message might appear that says, “Customers like you have been very successful using our bot. Would you like to try it?” If she wants, she can opt in, if not, she can choose chat or the channel of her choice.
If a customer has a history as a repeat caller, or appears to require a little more handholding, that person can be connected with a different agent group with the skillset to handle that type of customer. The goal is to take the time on the first interaction to completely solve the problem, and take steps to ensure that the customer is confident in the resolution.
On the other hand, if the customer is an enterprise client whose issues tend to be at a higher level, but solved very quickly via chat, perhaps that person is bumped up in the queue, so he can get in, get the issue resolved quickly, and get back to business.
By fully utilizing the customer data they have, and applying AI and machine learning, organizations can proactively identify how to provide the best experience to each individual customer. In doing this, companies are not just improving satisfaction levels but moving customer support from a reactive to a predictive model.
“In the past, customer service was viewed as something that was a reaction. Organizations were only benefitting when they were able to recover,” Melia said. “Now, organizations are learning to anticipate and deliver on customer needs and expectations, and by doing so, are creating deep brand loyalty that not only drives additional business with those customers, but makes them fans who will bring more customers and greater success to that brand.”
Instead of the traditional “I don’t know who you are, I don’t know anything about you, but I’m going to gather up all the information I can and figure out how to service you” approach, AI and data analytics help companies move to a model of “I know who you are; I know what you need, and I know how to give you the best experience.”
According to a recent article in CIO, using an AI-based system to connect the dots—aggregating current and historical caller information, and presenting the agent with the most likely reason for contact, makes the agents smarter and more likely to resolve the issue more quickly.
Taking Care of Issues Before the Customer Calls
As the AI and machine learning models mature, organizations have the opportunity to not only predict customer behaviors, but to start predicting issues that could prompt the need for customer support and take action before the customer ever makes contact.
For example, if an individual with a delayed flight contacts an airline, the virtual assistant would not only determine if that’s the reason for the call, but offer to book that traveler on a different flight.
Companies can proactively reach out to customers who are in danger of churn with a special renewal offer, before they cancel their account.
Insurance companies can reach out to high-risk individuals to refill prescriptions for them before they run out and as a result, need emergency care.
As more data is collected and analyzed, and machines accurately identify patterns faster, the more organizations can recognize triggers of certain events before they happen. Ultimately, this predictive capability will enable organizations to anticipate and act on a customer need—often before that customer even knows that the need exists.
Tempering the Tough Job of Content Moderation
In addition to providing support to customers needing help, many customer care organizations are also responsible for social media content moderation—ensuring posts with hate speech, bullying, violence and obscenity don’t make it to the public forum.
Essentially, a team of agents sifts through images, videos and text posts and eliminates anything that falls outside of the company’s guidelines. By anyone’s standards, it’s a difficult and potentially distressful job.
So, why not just train machines to take over and make the judgment call instead of human agents, and keep people from having to view or read disturbing content every day?
Although AI can learn quickly and well, the challenge with using AI as the sole judge of whether or not content is allowable or not is context. If you describe a cat and a lion, you might say they are both furry, with claws, tails and pointy teeth. But, one is a pet and the other one will kill you.
A word that in one context might be derogatory in another context could be considered funny.
“Automation could probably do that first step of identifying anything that might be hate speech or fake news or pornography, and then, send that on to a person or two people do the final check,” Mooney said. “It’s a similar approach to how ETS (Education Testing Service) grades essays. They used to have two teachers looking at and scoring the essay, and if they disagreed, it went to a third teacher. Now, they have a teacher and an algorithm scoring the essay. If they agree, all good. If they don’t give the same grade, then it goes to another teacher. The process is partially automated, but the human makes the ultimate judgment call.”
Companies first have to determine what they stand for; and what they want their policies to be around allowable content. For example, a more conservative organization may ban the use of certain words completely, whereas another may flag them for contextual review.
Again, how the company integrates AI into the process will depend on the level of risk—both the risk of something that should have been blocked getting out to the public, and the risk of censoring something that should have been allowable.
Getting Agents Onboard with the Change
To this point, we’ve talked primarily about the impact of intelligent technologies on the operational aspects of customer support. But, no matter how far AI and machine learning have come, the human agent remains a vital part of the customer support ecosystem. As such, it’s imperative that any support organization that plans on implementing intelligent technologies spend time getting its human workforce ready for, and excited about, the change.
According to a recent article in CMS Wire, organizations experiencing the smoothest transitions are those that start education early and tailor it to different stakeholder groups before anything is in place.
It all starts with communication. It’s important for your existing staff to understand what a machine can and cannot do, how the change benefits the customer—and, most importantly, why their capabilities are essential to the continued success of the organization.
Provide a forum through which agents can ask questions anonymously, in a safe space. Then, answer these at regularly scheduled town hall-style meetings.
Talk to individuals about how their jobs will change, and consider changing titles or job grades, if they’re taking on more or elevated responsibilities. Also, discuss their career path in the new environment.
The more that organizations can be transparent and open, and get the human agents involved in the change, the more they can quell staff anxiety.
“The whole reason any of us get into this field (AI) is to make life better for people. You want to get robots to do things that are dirty, dangerous and dull—to replace the work that people don’t enjoy doing, so they can spend time on the parts of their job that are more fulfilling,” Mooney said. “I think people need to have more of that attitude; to understand the real purpose of the technology and the positive impact it has the potential to make.”
Finally, actively invest in your agents’ mental health. Studies show that individuals can be taught to be resilient, which, in turn, enables them to become more adaptable to change. In an industry—and a world—where technology is continually evolving, and disruption has morphed into the norm, that adaptability is crucial.
By giving your staff these tools, you increase performance, job satisfaction and retention—all foundational elements of a strong customer support operation.
Recognizing AI-empowered Customer Support as a Comprehensive Business Strategy
Although providing an outstanding customer experience is essential to success, it’s not the only benefit of an AI-empowered customer support operation. Although smart companies have long used collected customer data to continually improve service levels, companies are now using this insight to guide everything that they do.
“It’s no longer just agents analyzing voice-of-the-customer data anymore. That information has to be integrated and leveraged throughout the organization—impacting how products and services are designed, how they’re packaged, how companies operate and market,” Melia said. “Organizations are increasingly using AI to analyze those customer interactions and look for patterns. They’re going from hearing the voice of one customer to seeing patterns across hundreds of thousands of customers. So, they can view the business landscape through a more comprehensive lens.”
No question, the customer support operation is fundamentally changing, powered by a more demanding consumer, the application of intelligent technologies, and a realization that the customer experience is a true business strategy.
“The customer experience has emerged from ‘the service center’ to ‘front and center’ for progressive organizations. As such, the expectations continue to get higher,” Melia said. “Although we don’t know what artificial intelligence in the future is going to look like, we do know that by looking at ways to use AI to supplement the work of their agents and staff today, companies can ensure they’re positioned to serve a global, diversified and increasingly demanding marketplace tomorrow.”
And that’s a competitive advantage.
For more information on AI & customer support, contact us here.
TaskUs is the fastest growing tech-enabled business services company in the world. We exist to provide ridiculously innovative customer support, AI operations, and content security to the world’s most disruptive companies.
Get on the list.
Dallas Innovates, every day.
Sign up to keep your eye on what’s new and next in Dallas-Fort Worth, every day.