One of the major ways business owners have leveraged AI and machine learning to their advantage is by using automated outreach for customer service. We’re all familiar with automated outreach, whether we know the term or not — we’ve all received those emails from companies or individuals reminding us we have an unpurchased item in our shopping cart, or asking for a review of a product we bought.
But automated outreach has gone even further than that when it comes to customer service. Many businesses now use AI chatbots to supplement their customer relations, which eases the need for human customer service reps while still maintaining high quality service.
How Do Chatbots Work?
So what are chatbots? A chatbot is a type of artificial intelligence (AI) that can carry on conversations much like a human does. It can interact with users in a conversational and friendly way, allowing customers to get basic or common problems solved quickly and efficiently, without having to involve a phone tree or email. They’re a great supplement to existing customer service, as they reduce the workload taken on by human agents, who can be reserved for more complex or difficult customer service issues.
Chatbots have their roots as far back as the 1950s, when Alan Turing introduced the “Turing Test ” — the question of whether a person could tell if they were talking to a machine or not. The first chatbot was created in 1966 by Joseph Weizenbaum. It was called ELIZA and while she could fool some users into thinking she was a real person, she didn’t completely pass the test. Online bots started emerging as early as 1995, with the famous A.L.I.C.E. bot. A.L.I.C.E. did pass the Turing Test and was the most lauded and advanced bot of her time. Since 2010, chatbots have become more and more common, with software like Siri, Google Assistant, Alexa, and Cortana.
But crafting a great customer service experience with a chatbot isn’t as simple as installing some software and calling it a day. An AI still has to be taught, and that means writing quality conversational scripts for the AI to use.
Designing the Chatbot Experience
That’s not to say you need to be a programming wiz to implement a chatbot — you don’t. Designing how the chatbot will interact with people involves conversation flows, which are a series of scripts and decision trees. With most chatbot software, you can either work from an existing template, or roll your own (once you’ve gotten the hang of it). Like a flowchart, decision trees tell the bot how to react, and what condition needs to be met in order to react that way.
To get started with your chatbot, the first thing you should do is decide what the goal of the chatbot will be. Is it meant to show customers around your website? Assist them with technical issues? Point them to other resources that can solve your problem?
Once you have your goal nailed down, you should make a quick sketch of how you imagine this conversational flow happening. You can do this on paper, with a tool like Google Drawings, or whatever works best for you. Don’t worry too much about being specific or detailed — the flow of conversation is likely to change as you work out your decision tree.
Once you have that tree fully sketched out, go back and polish up the dialogue, then do some pre-testing. From there, it’s a matter of simplifying and streamlining the experience as much as possible. After that comes a cycle of testing the prototype experience, refining, and testing again until you’re happy with the results.
As you might expect, the key to creating a friendly and believable chatbot experience means making the dialogue options as convincing as possible. The main ways you can do this include:
- Dialogue style and tone. Ideally, you want the AI to be friendly, assertive, knowledgeable, and inviting, guiding users forward without seeming pushy or invasive.
- Using conversational language. Because you want your bot to connect with users, it’s best to use an active, friendly voice that will establish trust while making the user feel like they’re in good hands.
- What terminology and jargon to use. If you’re working in an industry with a lot of jargon, you’ll have to decide how much (or how little) you want to include. You want to make sure users trust the AI knows what it’s talking about, while not confusing or intimidating them.
- Including just enough personalization. Personalization has become a major part of the modern customer service experience, because it’s such an effective tool. Referring to users by their first name when they first interact with a chatbot can greatly help with connection — be careful not to use too much, though, as users might get uneasy if you use too much personalized information.
Used correctly, chatbots can be a powerful addition to your customer service toolbox — as with any such tool, it all comes down to creating a positive, memorable experience.