When done right, chatbots can act as a concierge for your website, guiding your users on a journey toward helpful resources.
When done right, chatbots can also be used as a consumer support tool, helping your customers get answers or resolve problems.
When done right, chatbots can capture leads and generate revenue.
When done right.
When they’re done wrong, it’s nothing but a bad user experience.
After a lot of trial and error for ourselves, we figured out some successful applications for clients. We think it’s important to be honest about our own journey, so we’ll share some of that below, along with our chatbot success stories and how other companies are getting it right.
We started our chatbot journey with the goal of gaining conversions because we learned through testing that not everyone wants to fill out a static form on a webpage. The idea was to use artificial intelligence as a conversion point that gave our audience a product in exchange for their information.
Our first stab at bots didn’t do any of that. When we turned ours on, it wasn’t automated in any way — it was just sitting there waiting for chats.
The problem with that? We missed a lot of messages, which made for a bad customer experience.
If you’re going to have a live chatbot, you need to have someone ready to answer chats all the time. Obviously, right? Don’t turn on a chatbot for the sake of having a chatbot.
We went back to the drawing board and created a bot that asked some automated questions with the aim of making the chat experience more authentic. It also helped us gather lead info from the get-go, rather leaving the whole conversation open-ended and unanswered (but there was still the opportunity to chat with a live person at the end if the bot couldn’t answer a visitor’s question).
The problem? We used the same chatbot on all of our pages and we still missed a lot of messages.
The user experience on every page is different, so the chatbot needs to be different, too. A chatbot specializing in SEO help isn’t going to work on a page about branding. When you build a chatbot, it’s important to establish the bot’s function so you can set up that expectation with your visitor.
An open-ended chatbot, for example, lets people go into the experience with their own expectations (that they’ll get the help they want), and they’ll end up annoyed if they don’t get that experience.
Also, use bots sparingly at first to make sure you’re using them for the right reasons and not as an unintended way to drive people away from your site.
For this iteration, we kept the automated messenger bot, but instead of leaving it open-ended, we gave visitors the opportunity to schedule a meeting with a member of our team. This was one of the more successful iterations of the bot, but there were still issues.
The problem? Not everyone is ready to buy.
This approach doesn’t work for top-of-the-funnel visitors. If someone is coming to your site for the first time and wants answers to general questions, they may be turned off at the idea of talking with someone who is going to try to sell them a product or service.
We’re currently re-strategizing our chatbot strategy, but in the meantime, we’ve come up with some innovative ideas for our clients.
Despite our struggles, we have seen successful chatbot technology at work.
In a successful client application, we developed a chatbot for lawn care provider WeedPro. Its bot, which lives on a pay-per-click (PPC) landing page, can do one of two things: Estimate the price of lawn care or set up recurring lawn care appointments for current customers. Both of these experiences are automated.
This bot has a simple, straightforward purpose stated at the beginning of the interaction, setting the visitor’s expectation for the way the rest of the conversation will go. WeedPro already had a PPC landing page with a static form that wasn’t seeing the conversions company leaders had anticipated. With the addition of the chatbot, conversions increased from 20% of clicks to as high as 80%.In a newer application, Safe-T-Cover implemented a chatbot to help users find the outdoor utility pump enclosure they need. On STC’s PPC landing page, users have the option of getting a quote for a standard or custom enclosure. The Standard Enclosure option will open the chatbot, while Custom Enclosure opens a static form. The bot will ask the user for measurements and then suggest pump enclosure options for purchase. It will even send the visitor a spec sheet PDF if requested. This bot is still in its infancy, but we’ve seen great results. In an industry where a chatbot isn’t the conventional way of connecting with a company, the client has been happy with the preliminary results the bot has garnered so far.
There are also great examples marketers are looking to. Amtrak, for example, is using a chatbot named Julie to answer frequently asked questions and to help users book travel. Julie has been around since 2012 and saved Amtrak $1 million in customer service expenses in one year. It answers roughly five million questions each year and bookings through Julie generate 30% more revenue.
Insurance company Allstate uses a chatbot to help reduce calls to its call centers, therefore dramatically reducing call wait time. Dubbed ABIE, pronounced “Abby,” it can be found in the business insurance section of Allstate.com. It answers questions in real-time about small business needs and insurance solutions. Allstate claims ABIE answers simple questions like, “What is a deductible?” and more complicated questions like explaining how the claims process works. And because it’s artificial intelligence, it learns as users ask it more questions using natural language processing.
Don’t ever just turn a chatbot on. Remember open-ended chats invite everyone to chime in, including people trying to sell you products and spam. Use your chatbots to guide visitors toward the end result they’re looking for as quickly and efficiently as possible.
In a world where people are getting less and less patient, chatbots are a great way to fill a customer service gap. But don’t use a chatbot just because it’s what everyone else is doing. Getting it right will benefit you and your bottom line. Chatbots won’t work unless you do.
Chatbot development is tricky, and getting it right will take a lot of testing. Let us know what worked for you, or tell us your struggles. We’d love to hear about your trials and successes in the comments below.