TACTICAL TECH
Everything you need to know about a potentially powerful new communication tool
Chat bots have been around for a while. Exactly what are they?
According to Google’s definition, a chat bot is “a computer or program designed to simulate conversation with human users, especially over the Internet.” Chat bots can be used in any business, and we recently have seen an uptick in interest within the insurance space.
Let’s take a step back. In the early 2000s, chat bots were still in a beta phase, and much testing was required to develop bots with a higher “IQ” or overall range of functionality and ability to understand conversational flows. A popular bot that was open to the public for chatting during this time was Cleverbot, which is still available online. If you’d like to get hands-on experience with a chat bot that has a 10-year history, check out cleverbot.com.
Some chat bots are set up to learn, such as Cleverbot, but most are manually programmed by a developer through the back end. This means that to program your bot to understand more, you need to log into the bot’s hosting service and add to its intelligence, one command at a time. If your agency decided to implement a bot on your website or on social media like Facebook Messenger, you could extend the limits of its functionality through the bot’s back end by programming it to understand more, but this requires a watchful eye and ongoing maintenance.
Let’s say you wanted to program a chat bot for your website that could receive quote requests from your clients, allow them to get in touch with your agency, and enable them to get help with claims. First you’d implement your bot using a service such as ProNavigator, a company that specializes in building bots for insurance agencies, which you can find online at pronavigator.io.
Next you’d work with a team of chat bot professionals to program your agency’s bot to service your clients and prospects how you’d like. At this point, your bot would have keywords or phrases that trigger specific conversations. For example, if someone opened up your bot after this programming, the conversation could look something like this:
Bot: “Hi, welcome to our agency. How can we help you today?”
User: “I need an auto insurance quote.”
Bot: “I see you’re looking for assistance with a quote for an auto insurance policy. I’ll just need some information from you first. Can we start with your first and last names, please?”
And so on. This bot recognized the phrase “auto insurance quote” and immediately initiated a conversation that specifically responded to the user’s request. Although this makes for a great emulated conversation, it’s important to remember that this is, in fact, an emulated conversation, meaning that the bot is only as smart as you programmed it to be. It’s following the patterns of an actual conversation, but it can only process specific keywords or phrases and trigger responses accordingly.
This means that if a user types in something that is outside of what you’ve programmed the bot to understand, he or she will not be able to be serviced by your bot. The bot in our example has been programmed to help with quotes, claims, and getting in touch with your agency. What if a client wanted to visit your website to post a review of your agency? Here’s what the conversation would look like:
Bot: “Hi, welcome to our agency. How can we help you today?”
User: “Leave a review.”
Bot: “I’m sorry, but I can’t seem to figure out what you’re asking. Do you need help with anything else?”
User: “I’d like to leave a review.”
Bot: “I’m sorry, but I’m still having trouble figuring out your request. Would you like to have one of our professional agents give you a call?”
When the user makes a request that is outside of the chat bot’s functionality or programming, the responses can be clunky or unexpected. The user is searching for help with leaving a review, but the bot can’t deliver, because it doesn’t recognize any of its programmed keywords or phrases and, therefore, cannot take action.
This sort of behavior by bots, which prevents the user’s issue from being resolved, has caused bots to have an alarming bounce rate. Another kind of clunky conversation occurs when the bot cannot interpret the user’s text. A user may make a typo or send a single statement through multiple messages. Here’s an example, first with a set of split-up messages:
Bot: “Hi, welcome to our agency. How can we help you today?”
User: “Hi, I need a quote”
User: “for my renter’s insurance”
Bot: “I’m sorry, but I can’t seem to figure out what you’re asking. Do you need help with anything else?”
The bot can’t immediately recognize messages that are split up, and usually the conversation will play out like the example above when a user unintentionally splits a message into multiple parts. Although bots are being developed to be “smarter” and figure out situations like this one, they aren’t 100% ready yet.
Let’s look at another example with some simple typos:
Bot: “Hi, welcome to our agency. How can we help you today?”
User: “Hi, I need a quote for my retners insuracne.”
Bot: “I’m sorry, but I can’t seem to figure out what you’re asking. Do you need help with anything else?”
Did you notice that renters insurance was misspelled? The bot certainly did. In fact, if a typo is made in a certain word or phrase that normally triggers conversation from the chat bot, in most cases it won’t know what to do. These kinds of clunky conversations cause users to bounce away from bots because they didn’t get the service they needed. In fact, a friend from Forrester has told us that when bots don’t provide the experience the user is looking for, the user is apt to bounce away from the bot more than 60% of the time. That’s a surprising percentage and indicates that, overall, bots aren’t quite ready yet to provide an outstanding customer experience.
What does all this mean for you and your agency? For Paradiso Insurance, it means that we see an exciting future ahead of us in terms of communicating with our customers, but we are going to wait for the right time to integrate a chat bot into our website. Conceptually, the chat bot is a great tool and can be a valuable asset for any agency, and we’ll begin to use one after these bots become smarter and able to answer more questions from our users. In the meantime, let’s watch the behavior of bots, especially within our industry, to see how they will improve the customer experience going forward.
The author
Chris Paradiso, CPIA, is president of Paradiso Financial & Insurance Services, headquartered in Stafford Springs, Connecticut. His agency won PIA National’s Excellence in Social Media Award in 2013. He also heads up Paradiso Presents, LLC, which provides social media consulting, seminars and workshops to help agencies thrive in the online marketing world. Contact Chris via email at cparadiso@paradisoinsurance.com
