Is the Call Center Disappearing? LivePerson, IBM and ISG Think So

There exists an ideological tug-of-war between customers and businesses. While AI vendors generate ceaseless streams of research extolling the virtues of non-voice channels for speed, efficiency and operational cost savings, 71 percent of customers still prefer to talk on the phone with a live agent, our research shows.

Nonetheless, self-service infrastructure is emerging as an antidote to long wait times in phone queues, to the point where experts predict that by 2020, eighty-five percent of all customer interactions will be handled without a human agent.

With the rise of conversational commerce, some of the biggest brands offer full, AI-assisted customer support on instant messaging apps like Facebook Messenger, Whatsapp and Apple Business Chat.

Given the level of automation and convenience customers are primed to expect, it’s little wonder that experts predict the demise of the manned contact center as we know it.

Some organizations - and their customers - are more digitally advanced than others. For an organization that still handles the majority of customer support interactions through the voice channel, a successful digital transformation has to move at the pace of the customer.

“You have to think about where your customers are at the moment,” Wayne Butterfield, Principal at ISG, a leading global technology research and advisory firm, said at the recent Future of the Call Center Virtual Summit at IBM Watson headquarters in New York City. “So if you’re a very voice-centric organization, to get your customers from the phone to a digital channel is a journey itself.”

By starting with the channel customers predominantly use, rather than abruptly introducing a new one, digital transformation becomes a matter of incrementally augmenting that channel through AI. This way, they have the opportunity to experience the speed and efficiency gains of digitization themselves and adjust on their own time, rather than the feeling of being left in the lurch when they pick up the phone and encounter for the first time a circuitous IVR clearly designed to deflect calls.

Voice-centric organizations should start with the IVR, said Rob LoCascio, founder and CEO of conversational commerce company LivePerson. It can be as simple as informing callers of an SMS customer support service, while also giving them the option to hold the line to speak with a live agent. The customers that do convert should then be handled by a live agent. In the meantime, the organization can build transcripts of those interactions and use those transcripts to train the bot.

“Then look at the intents, script the questions and answers, and then have your agents watch the bot,” said LoCascio. Once the bot is sufficiently trained, let it perform an end-to-end interaction with a customer while using sentiment analysis monitoring and have agents observe it in real-time in case they need to intervene.

“Once you get past that, starting small is not the worst thing in the world.,” Butterfield concurred. “Both the major deployments of Watson and LivePerson in the UK at Royal Bank of Scotland and Vodafone have started off reasonably small.”

A gradual approach is also key to ensuring you don’t roll out a half-baked AI solution that hasn’t fully familiarized itself with a sufficient range of scenarios and queries to be unleashed to the real world. Butterfield, who has over 10 years of experience consulting with companies on leveraging new technologies to become more efficient, says that a realistic AI deployment takes about three months.

“This is about creating an experience that is as good if not better than what the consumer is used to in their usable channel of choice,” Butterfield warned. “That is the only way that you will create long-term sustainable channels.”

Start with a simple solution that doesn’t require integration with any back-end systems like a CRM tool. “As soon as you start getting into integration, you’re talking about involving many parts of your organization.”

When UK-based telco Vodafone first implemented AI, it did just that. The company worked with IBM to build a virtual assistant named TOBi, who could handle routine customer queries, such as asking about data roaming charges before traveling abroad. Eventually, TOBi was trained to access specific customer data on the back-end to identify if they had a roaming bar in place, authenticate the customer, and complete the task of taking away or adding a roaming bar.

“What we’ve found is that the low-value, almost boring transactional-type activities are being taken away from them as a human and ultimately seeing that they are getting retrained and refocused towards case management and those types of activities,” said Neil Blagden, COO for Vodafone UK.

According to LoCascio, 268 billion calls are made to customer service hotlines each year, at a cost of $1.2 trillion. By comparison $500 billion is spent on advertising globally, while e-commerce spending totals $2 trillion. “In the middle of e-commerce and advertising globally is this thing called analog voice calls. And it needs to get digitized.”

For Saas businesses, automating customer support isn’t necessarily an option, especially for those built on a subscription model. In fact, they need more robust customer support systems to deal with everything from level 1 support tickets on billing inquiries and general confusion to technical troubleshooting by qualified customer support engineers.

When Autodesk, provider of 3D design, engineering and automation software built its first virtual assistant, it was also a matter of automating routine queries in support tickets, and using the AI to sort and escalate the more technical ones. The company shaved resolution time from 1.5 days where agents were consistently tackling backlogged, repetitive queries to just five minutes - amounting to a 99 percent faster resolution time.

“If nothing else, if we can just understand what customers want, we can route more appropriately, collect more information and create a case so that when it gets to a human agent, they’re not having to do all that work,” said Gregg Spratto, VP of operations at Autodesk. “This ultimately leads to quicker resolution and a better customer experience.