by

Chatbots have been around for a while now, whether you’ve heard of them or not. What are chatbots exactly though? If we look at Google’s exact definition, it states that chatbots are “A computer or program designed to simulate conversation with human users, especially over the Internet.” Yes, chatbots are in place to serve one purpose, and that is to emulate a conversation with users who visit you online, such as on social media or on your website. Bots can be implemented into any business, but we recently have seen an uptick in interest within the insurance space.

Actually, let’s take a step back –  Did you know that back in 2004, there was a spike of interest with chatbots across the web, because this concept was relatively new during that time period. I’m sure that’s an eye-opener for some of us, especially if we are just hearing about bots for the first time. That being said, back in the early 2000s, chat bots were still in a beta phase, and much testing needed to be done in order for bots being developed to have a larger “IQ” so to speak, or overall range of functionality and understanding conversational flows. One of the bots that was more popular that was open to the public for chatting during this time period was Clever Bot, which is still available online. If you’d like to get your first hands-on experience with a chatbot that has had roughly ten years to develop, I would check out cleverbot.com for a quick moment.

Now, here’s the thing; some bots are set up to learn, such as Clever Bot, but most bots are manually programmed by a developer through their back end. This means that to program your bot to understand more, you’d have to log into the bot’s hosting service, and actually add to its intelligence, one command at a time. For instance, if your insurance agency decided that you wanted to implement a bot on your website, or even on social media such as through Facebook messenger, the bot would be limited on its functionality based on how you set it up. You could push the limits of its functionality any time through the bot’s back end by programming it to understand more, but this could take a careful eye and ongoing maintenance. Allow me to explain further.

Let’s say you wanted to program a chatbot for your website that could receive quote requests from your customers, get in touch with your agency, and 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 that you can find online at https://pronavigator.io. After that, you’d work with a team of chat bot professionals to help program your insurance agency’s chatbot to service your customers in the ways that you listed above. 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 would look something like this:

Bot: “Hi, welcome to our insurance 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 name please?”

And so on. This bot recognized the phrase “auto insurance quote,” and immediately started initiating a conversation that specifically entailed to the users request. Pretty intuitive right? While this makes for a great emulated conversation, the important thing to remember here is 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, then they will not be able to be serviced by your bot. For instance, we mentioned that the bot in our example has been programmed to help with quotes, claims, and getting in touch with your agency. Now, what if a customer wanted to come to your website to leave a review for your agency? Here’s what the conversation would look like:

Bot: “Hi, welcome to our insurance agency. How can we help you today?”

User: “Leave a review”

Bot: “I’m sorry, 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?”

In these instances where the user requests something that is outside of the chat bot’s functionality or programming, the responses can be a bit clunky or unexpected. The user is just trying to search for help leaving a review, but the bot can’t deliver because it doesn’t recognize any of its programmed key words or phrases, and therefore, it cannot take action.

This sort of behavior from bots, which causes the original “problem” of the user to remain unresolved, has caused bots to have an alarming bounce rate. In fact, there’s one other type of “clunky” conversation that exists far too often with bots, which lies in a problem of interpretation of the user’s text. Sometimes, users may have typos, or even send one single statement through multiple messages that should have all been on one. Let’s take a look at an example, first with a set of split up messages:

Bot: “Hi, welcome to our insurance agency. How can we help you today?”

User: “Hi I need a quote”

User: “for my renter’s insurance”

Bot: “I’m sorry, 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 in all scenarios, and in fact, most of the time it will play out like the example above when a message gets split into multiple parts unintentionally by the user. While bots are being developed to be “smarter” and figure out situations like this one, they aren’t 100% ready yet. Let’s take a look at another example with a simple typo instead:

Bot: “Hi, welcome to our insurance agency. How can we help you today?”

User: “Hi I need a quote for my retners insuarnce”

Bot: “I’m sorry, I can’t seem to figure out what you’re asking. Do you need help with anything else?”

Did you notice that renter’s insurance was misspelled? The bot certainly did. In fact, if there’s a typo on a certain word or phrase that normally triggers conversation from the chatbot, it won’t know what to do at all in most cases. These sort of “clunky” conversations that we keep mentioning have caused users to bounce away from bots, because they didn’t get the service they needed. In fact, one of our friends 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 amount, which means that overall, bots aren’t quite ready yet to handle providing an outstanding customer experience, which is something our agency focuses on.

So, what does all this mean for you and your insurance agency? Well, for Paradiso Insurance, it means that we have an exciting future ahead of us with communicating with our customers, but we are going to be patient and wait for the right time to implement a chatbot into our website. The chatbot, conceptually, is a great thing to have, and can be a valuable asset for insurance agencies everywhere. We believe that chatbots need more time and development to be smarter and answer more questions and recognize more problems with communications with our users. We just believe that they aren’t quite ready yet. In the meantime, let’s watch the behavior of these chatbots, especially within our industry, to see how they end up changing the customer experience moving forward.