Graeme Dawes - Fotolia
- Bridget Botelho, Editorial Director, News
AI for customer service tools allow mega corporations to use all the data they collect to better understand consumer behavior and derive predictive insights to boost sales, while theoretically improving the customer experience by delivering personalized recommendations. The irony, of course, is that these hyper-personalization tools also serve to prevent customers from having personal interactions with the people at those companies.
But that's the way of the world today. We're so accustomed to getting what we need from automated systems -- be it our preference to use the ticketing kiosk at the airport instead of speaking with an agent at the counter or the ATM at the bank we rely on in lieu of a teller -- that many of us only seek out a person when automation fails.
I, for one, will pick up the phone and talk to a customer service agent only as a last resort. I prefer to find solutions to product problems using self-service platforms or instant messaging chats, and I don't really care whether the agent on the other end of that platform is live or virtual as long as they resolve the issue quickly. When I do take the old-fashioned telephone route, I expect to be frustrated by Interactive Voice Response systems that lead me on a circuitous route to my final destination, only to hear that I should visit the online help center because of larger than normal call volume increasing the wait time.
I'm not alone in my impatience and expectations. Gartner analyst Michael Maoz led a webinar earlier this year on how artificial intelligence affects the customer experience and how AI for customer service changes the way companies operate. He said nothing annoys customers more than service agents who don't know why they're reaching out in the first place -- and I concur. By the time customers reach a live agent, it's likely they've already spent time trying to solve their issue on other channels.
Serve them on location
Companies are also expected to provide products and services across platforms to meet customers wherever they are. Today's consumers might see something on Instagram or Pinterest, and their path to finding the product to purchase might be through an app, a social media channel, the company's website or the store. Retailers that once relied on the "If you build it, they will come" mantra now need to build it, locate their customers and serve them there.
"We have to put ourselves in the shoes of the customer and find out how they expect things to get done," Maoz said in the webinar. "Consumers expect things to be as easy as possible. They expect what they need to show up on whatever channel they're using."
Consumers didn't develop these expectations overnight; they've been conditioned to expect as a matter of course personalized service by the companies that serve it up in the same way we have all come to expect Wi-Fi access everywhere we go. Hell hath no fury like a business traveler who can't get Wi-Fi on an 8-hour flight.
When service levels increase, so will customer expectations, along with greater reliance on AI for hyper-personalized customer service. AI strategies are a top priority for executives, and the need for artificial intelligence is a big factor in their technology decisions.
Gartner predicted 72% of customer interactions will involve machine learning apps, chatbots or mobile messaging by 2022 -- up from just 11% of interactions last year. And in three years, 15% of all customer service interactions will be handled completely by AI. Not surprisingly, fewer and fewer people will actually pick up their phone to call customer support lines; Gartner estimated that by 2022 only 12% of customer interactions will be by phone, down from about 41% today.
No pain, no gain
Anyone who has been frustrated by automated customer service may cringe at these stats, but as Maoz said in his webinar, we're still in the experimental phase. Companies are learning and transforming the way they deliver customer service. At this phase of the cycle, disappointment is inevitable. But as companies move through their growing pains, they'll find massive value in AI for customer service -- and chatbots are only the beginning.
Image recognition isn't widely used yet, but its potential for customer sentiment and behavior analysis is great. Toyota and other carmakers, for example, are experimenting with image recognition to predict driver behaviors.
While AI tools are good at identifying what customers need and completing simple tasks, human agents are still very much necessary and will continue to be when it comes to providing support for complicated issues. Customer questions that require a lot of analysis is an area where customer service agent jobs will be secure.
"You need to keep a human in the loop, so when there's a problem the system can't handle, there is a person there," William Mark, president of information and computing sciences at SRI International, told attendees at last December's AI World in Boston.
For simple processes, AI and self-service will be the default options. And though the toughest tasks will remain with live agents, AI for customer service tools will allow those agents to improve the way they work. Speech analytics tools can enhance future interactions, and predictive analytics tools can assist with decisions about next steps. These and other technologies are not only agent-enabling, but also customer-enabling in that they help both sides get what they need from one-on-one interactions, Maoz explained.
The combination of humans and AI for customer service will continue to be the most effective approach. And though we rely on live agents less and less, it's good to know we'll have a lifeline when we need one.
- Marketing analytics, big data and ethics –TechTarget ComputerWeekly.com
- MicroScope – July 2023: The pressure to deliver –MicroScope
- 2021 UKI Media Consumption Infographic: Engaging Prospect to accelerate the buy... –TechTarget ComputerWeekly.com
- 2021 UKI Media Consumption Infographic: Digital research preferences –TechTarget ComputerWeekly.com