IBM’s Watson computer system defeated legendary game-show champions on Jeopardy by combining a deep database of knowledge with the ability to quickly process natural language questions, and provide quick natural language answers. (Well, it was Jeopardy, so it actually was listening to answers and providing questions, but you get the idea.)
Chatbots aim to do the same thing on a more practical level. They provide the authentic experience of talking with a person to get answers and conduct transactions, without actually having a person on the other side of the conversation.
The natural language element of this advance in artificial intelligence development is crucial. For most of the history of computing, people have needed to adapt to a special interface in order to tell computers what they want and how they want it. Natural language processing lets chatbots understand what people want by listening and responding as a person would. They can understand orders, record payment information, and process transactions. They’re already so powerful that a BI Intelligence analysis projects they could replace enough sales professionals to save businesses $65 billion in annual salary.
And they’re becoming the next wave in digitally-driven sales for the same reason search engines are such powerful advertising platforms. When you get people talking about their needs, even in just a few words or a phrase, you learn a tremendous amount. Get them to start expressing themselves in full sentences, or engaging in a full dialogue, and you gain a tremendous advantage.
Why chatbots work
Chatbots have been entertaining people since they were created in the 1960s. ELIZA, the best-known early chatbot, could fake her way through a therapy session for a few minutes. People like communicating with chatbots for many of the same reasons they like texting and messaging their friends. It’s fun to dash off messages to friends and see what they say.
Unlike people, who have other friends to answer (not to mention lives of their own to lead) chatbots never get distracted and never sleep. They can be programmed to provide a near-instantaneous response to any message, at any time. That means more opportunities to get your customers talking about what they need and want, even when no one is awake to listen. Chatbots designed by Fiverr seller michaellam924 can even take survey answers from customers.
Chatbots are also a natural extension of the growing trend for companies to offer sales and service via Facebook, Twitter, and text messages with human agents. As people grow more comfortable with doing everything – from opening and closing accounts to ordering and returning products – through these real-time text communication windows, it will be easier to shift people to engage with a chatbot. The medium is the message, as the saying goes. Who or what is sitting on the other side of the conversation isn’t as crucial.
How chatbots help customers
Not all chatbots can take a customer through a complete buying process, but they don’t have to. Simply encouraging people to volunteer information, to start filling a shopping cart, or to ask a question gives you a great head start. You can send the session to a live person to continue the chat, offer to phone or email with more information, or store the discussion and offer to pick it up at a later date exactly where the prospect left off.
This works particularly well in high-value situations, like the market for legal services. Chatbots like Elexirr are cropping up to provide basic input on potential legal action, then make informed referrals to participating lawyers if the client wants to proceed. The handoff helps the lawyer access more potential clients and can save valuable time.
What to ask a chatbot
Although chatbots don’t have to apply for a job through the usual channels, you can still think about them like an employee in one regard: the interview. Start interviewing the chatbots you come across, asking them to solve real-world product-recommendation, customer-service, and sales needs. You’ll get a good feel for what’s possible today as well as where the remaining rough edges are. Make a purchase, and see if it’s an experience you would feel comfortable putting in front of your own customers. You may get a surprising response that helps you better understand how a chatbot can enhance your ability to communicate with customers 24/7.
What customer service problem would you like a chatbot to help your business with? Is there anything you would never trust to a chatbot? Tell us about it in the comments.