ON THE SCENE—Customer Service Experience: Bots Are Only as Good as the Data

The old adage that the customer is king no longer applies. Today’s customer is a ruthless dictator, exercising total and utter control over just about every aspect of any business interaction, speakers emphasized at this year’s Customer Service Experience conference in Washington in late April.

And because customers are so demanding, businesses need to respond with the right mix of channels and the intelligence to get the right information to the customer in the shortest amount of time possible. That was one of the potent messages of the three-day conference, which was collocated with CRM Evolution.

Artificial intelligence and natural language understanding are two of the main tools to help businesses meet those tough consumer mandates, but they are not easy to implement, due in large part to the speed with which the technologies have evolved, according to a number of speakers.

And the importance of natural language and artificial intelligence isn’t limited by channel or demographics. No matter the channel, the customer still wants to communicate naturally, without having to follow a preset script or guided dialogue.

Nowhere is this more obvious than in interactions with intelligent assistants, which seemed to be the technology on everyone’s mind this year. In fact, an entire keynote panel was dedicated to chatbots alone.

While the technology has come a long way, it still has a long way to go. “There’s a lot of hype around artificial intelligence, chatbots, virtual assistants, and other forms of automated self-service,” said Ian Jacobs, cochair of the conference and a principal analyst at Forrester Research. “I have not seen any yet that are too spectacular.”

Esteban Kolsky, founder of ThinkJar, said that chatbots are “very functional today,” but said the technology “needs to break through the barrier of cognition.”

Chatbots, he added, “are a great tool, but they’re still very simple.”

But that’s not to say that companies should wait to deploy them. Chatbots, Kolsky said, are great for collecting form information and delivering simple information, like bank balances or flight times. Where they have problems, he and others stressed, is in handling more complex interactions.

“The trick to chatbots is putting them within the right boundaries,” Kolsky said. “Limit the number of questions. The chatbot is there to answer one question with one answer.”

And when chatbots can’t find that one right answer, the whole interaction can go awry. That, according to Sheryl Kingstone, research director for business applications at 451 Research and cochair of Customer Service Experience, is more a fault of the data available to the bot than the bot itself. “If you can get the right content to the right person, there’s no problem with automation,” she said. “The problem, more often than not, is that the data is bad.”

And that can’t be allowed to continue, she stressed. “Customers will always have questions. Questions are not going away. All the chatbot is doing is taking some questions away from the agents.”

And while the tendency is to give chatbots clever names and personalities, that is not where the focus should be. “The chatbot doesn’t need a fancy name; it just needs to guide the customer to the right information,” Kingstone said.

That’s the end goal of any customer interaction, Kolsky added. And that’s why customers turn to so many different channels. “All the customer wants is an answer, and the only reason the customer goes to one channel is because he can’t find the answer on the other channels. If you do it right on one channel, customers will use that channel and not have to seek out other channels,” he said. “Stop focusing on the channel and start focusing on providing the right answer. The customer doesn’t care how the information is delivered as long as the information is right.”