Solvvy

Artificial Intelligence in Customer Service: The Past, Present, and Future

 | 

Getting the customer experience right is a delicate balance.

Sure, you might be able to check your bank account balance thanks to the friendly robot on the phone—but if you actually wanted to talk to a person about an issue with your account and you’re left on hold for five minutes while the recorded operator recites a menu without one opportunity to talk to a human, you might be ready to shift your paycheck elsewhere. 

97 percent of consumers said that a bad buying experience will lead them to change their buying behavior, according to a recent Zendesk study. But what are they looking for in a good customer support experience, and what role can artificial intelligence play in increasing customer satisfaction? 

To answer that, let’s take a look at how artificial intelligence has been used in the customer experience in recent years, where it’s failed, and where we see the future of machine learning in delivering higher customer satisfaction.

solvvy reduce repetition graphic

Replacing human customer service agents with IVR technology: Bad idea.

One of the earliest forms of AI customer service to come mainstream is interactive voice response (IVR) technology, in which a computer recognizes the customer’s natural speech or touch tone and uses it as a prompt for their next action. This technology has been in use in some form since the 1970s, and grown far more common in the past decade. The problem? Customers hate it. 

A recent Harris survey found that 94% of customers had been frustrated with voice channels for customer support, and 47% complained that they weren’t able to bypass an IVR system to talk to a customer service agent. They were upset because the IVR options didn’t include the proper use case. And, in some cases, the IVR system wasn’t hearing them correctly: 45% of respondents to a Vonage study said that they were often forced to repeat themselves.

In fact, more than half of customers have chosen to abandon a business altogether because they’ve reached an IVR.

Though voice-recognition technology has advanced over the years, it still makes for a sub-par customer experience.

Chatbots are the next big thing… or not.

It turns out, most customers don’t even want to resolve their support issues over the phone—they’d rather do so online.

As such, many businesses turned towards developing machine learning-driven messaging platforms known as “chatbots.” Facebook began offering Facebook Messenger “bots” to brands on their platform in 2016, and many other companies quickly followed suit. While many businesses expected that chatbots would be the future of customer support, it hasn’t really turned out that way. 

Why not? More often than not, chatbots aren’t good at answering users’ questions. In fact, the Facebook chatbots’ failure rate was reported at a dismal 70%.

Chatbots don’t provide a great user experience, either. Forrester found that many customers find that they still have to talk to a customer support agent after a chatbot conversation, which wasted their time. Two-thirds of customers are now skeptical of chatbots and their ability to provide a good customer experience. Particularly when they are given a name and presented as equivalent to a human customer support agent, they often fail to grasp the nuances of a real-time text discussion, leaving customers exasperated. 

For example, a now-defunct weather app called Hi Poncho struggled to grasp the meaning of conversational follow-up questions. It could answer a question like, “Do I need a jacket?”, but was clueless when it came to responding to “What about an umbrella?” as you can see from this reporter’s attempt at a chatbot conversation.

While many brands are still using chatbots, they haven’t been the game-changer for customer support that many businesses had anticipated. They tend to perform best when they’re used for clear, utilitarian functions, i.e. setting up an onboarding sequence for a new user. If you ask your chatbot to do too much, and show personality while doing it, you’re likely to end up with conversational fails and exasperated users.

Machine learning in self-service support

Finally, that brings us to the AI-driven self-service platform. 

Like chatbots, a self-service solution can perform natural language processing to surface answers to customer questions. But unlike chatbots, self-service solutions aren’t heavily scripted—a risk that often makes it difficult for a customer to find the right answer or even get access to a customer support agent. Instead, self-service platforms make it easy to help customers find the right information inside your knowledge base, and routes their message to a customer support agent at any time if the issue is not self-serviceable. Self-service customer support platforms can be used to showcase common questions directly in the chat interface, so the most common customer requests can be answered instantly. More complicated questions can be routed to in-depth written or video tutorials within the knowledge base, or directly to a support agent.

By putting the power into the customer’s hands, you can ensure they’re not spending time trying to resolve a problem with AI that it’s simply not suited to. Self-service allows them to choose whether the provided information answers their questions, and, if not, to contact a support agent. This approach streamlines the customer support process, resulting in a better customer experience.

solvvy improve support graphic

The future of customer service isn’t a fully automated world

A decade ago, many business leaders might have assumed that with the advances in artificial intelligence we’ve seen to this point, customers would never need to talk to a live support agent again.

And while that might be true in theory, it simply doesn’t make for a good customer experience—and it doesn’t make good business sense, either.

A great customer experience is a key brand differentiator. A Walker study found that 86% of customers are willing to pay more for a better customer experience.

That encompasses everything from their first interaction with your brand—whether an ad, a blog post, a trade show booth, or an in-store display—to your online purchasing experience, to your customer support experience. It doesn’t matter if you’ve spent millions on a cutting edge website and branding package—if your customers are treated as disposable the minute they need to ask someone for help, they’re likely to feel alienated from your brand and decide to move on to a competitor that does value them.

Instead, we’re much more likely to see brands build artificial intelligence-supported customer experiences that use machine learning to automate and streamline simple processes, but facilitate customer service interactions based on each user’s personalized needs. That could mean:

  • Growth in self-service solutions
    We’re likely to see more brands build robust internal knowledge bases, and use AI customer service tools to quickly guide their users to where they need to go—whether that’s answering a simple question using existing content in the knowledge base, or routing them immediately to a support agent for a complex issue. As such, they’ll be able to reduce their support ticket volume and the number of agents they need to hire, allowing them to focus on ensuring their staff is highly trained to handle a complex range of issues.
  • The death of email support
    Though 54% of customers use email as a support channel today, Gartner predicts that 85% of customer interactions can be handled with AI technology by the end of 2020. That means we’ll see more and more businesses turning away from email towards online chat and self-service options, which can route them to a phone number when necessary. Customers demand real-time resolution to their issues—brands that make their customers wait to resolve their support tickets will quickly fall by the wayside to competitors that focus on speedy response times.
  • The rise of omnichannel customer support
    Customers move between devices all the time, and they want to be able to access the same customer support on their mobile devices as on their desktop computers. Innovative customer service solutions should make it easy to maintain your support history and the same functionality while moving between devices and platforms. Make sure your artificial intelligence-driven customer support solution works seamlessly across all the operation systems that your customers are using.
  • Strong integrations with CRMs
    In order to create a positive customer experience, you need visibility into the entire customer journey. Incorporating an artificial intelligence customer support tool within your CRM helps you understand and respond to the entire context of the user’s interactions with your brand. Leading solutions like Solvvy provide integrations with Salesforce, Zendesk, Kustomer, Oracle, and many other companies, so they fit seamlessly into your existing tech stack.

solvvy make informed decisions graphic

Machine learning should optimize the user experience, not fully automate customer support

We predict that artificial intelligence will make it easier for brands to deliver best-in-class customer experiences—but not by alienating customers who still long for a human connection.

Customers who want to resolve problems independently will have the option to do so, with an array of self-service solutions to quickly guide them to a resolution from a company’s robust knowledge base. Over time, we’re likely to see even more opportunities for personalization in self-service—for instance, some solutions can already identify where a user is logging in from, and may customize its response to a search query around shipping costs and timing based on location. We’re likely to see even more interactions that can incorporate a user’s entire history of experience with the brand to provide more detailed and useful support, such as by using IoT data to understand exactly what issue the user is having with a product when she gets in touch for support.

Customers who have more complex needs that can’t be easily answered by a script should have the option to be automatically routed to a customer support agent—as should customers who simply prefer to talk to a human. In this case, we anticipate a trend of moving away from impersonal call centers, to highly trained, personable in-house agents who’ll instantly make their customers feel better. By reducing the volume of agents you need to hire with the help of technology, you’ll be able to focus on hiring best-in-class customer support agents who can differentiate your brand with their approach to customer happiness.

Artificial intelligence is a great addition to your customer service tech stack when it’s used to automate repetitive tasks while improving your customer experience, cutting down response times and providing quick access to customer service teams when needed. Chatbots and IVR technology have struggled because they aim to replace the need for human interaction, or make it more difficult to access a human agent. By focusing on AI systems that incorporate self-service customer support technology, your brand can enhance your customer experience without losing a human touch.