Lately, every time I visit a website on the internet, I’m greeted by this conversational agent called a chatbot. Sometimes it takes on the disguise of a human like Flo or Jack. Other times the chatbot looks like a robot and reminds me it’s not a human. But in both cases, that bubble is engaging me to help with whatever intent I may have for coming to the website. That’s certainly helpful and has changed the way I interact with companies who traditionally had no digital face for me to connect with. But it got me thinking… what if I could replicate a chatbot for my own use? Say I had my own product or service I wanted to offer and as an additional channel for an engagement, I could hire a chatbot to work on my behalf? This sounded like a great idea and I was excited to dive in. Today I’d like to share some of my findings that may be surprising for you in your search for implementing your first chatbot.
Chatbots are bigger than customer serviceI always thought the only way chatbots were being used was to supply customer service. People would have questions and the bot would reply with an answer. This simplistic model for conversation is certainly useful at augmenting the high volume of inquiries coming in daily. Take your local bank or retail establishment. People want to know the hours of operation and refund policies. They want to know what impact crises like COVID-19 have on the business. And of course, they want to complete transactions whether it be ecommerce purchases or fund transfers. Chatbots are doing all of these things today in the mainstream, and there’s been an expansion in other sectors for chat usage. Sales and marketing professionals are generating leads. Recruiters are gathering candidates for job openings. Human resource members are addressing internal employee needs with an augmented chat team. Chatbot usage is available anywhere conversations happen and that is magical thing. Here is a list of some of the most common use cases available. Can you think of any that aren’t on the list? Can you think of any use cases chatbots could have not listed on here?
Great chat is driven by amazing user experiencesChatbots today are powered with artificial intelligence using a standard framework called Natural Language Processing (NLP). It’s quite an extraordinary field of study. But most people don’t have the interest to learn coding languages like Python to build new chatbots from scratch. Fortunately, many amazing developers have created intuitive platforms that speak the everyday human language. This allows you to focus on creating a great experience for the user interacting with your chatbot. Instead of spending your time coding, now you are mapping user journey’s on flowcharts and crafting captivating copy, intricate graphics, and a stunning avatar personality to delight their chat experience. Creating a flowchart to map your intended user journey shed clarity on how to build your chatbot. UX for the win! The option to create a chatbot from scratch is still available, but when pre-built frameworks like Bluefish.AI are already available, it’s a no-brainer to focus your energy on the most important thing—your end user.
Chatbots need loving care and attentionThere’s a well-known story in the chatbot community about a chatbot that was unleashed into the Twitterverse to be an autonomous agent. Developed by Microsoft, Tay was supposed to be the ultimate brand ambassador for its business master. What followed was one of the worst disasters a chatbot owner could have. Tay adapted to its human users by developing a racist, pro-nazi personality that dealt more harm than good. Not something that reflected well for Microsoft’s brand image. When you develop a chatbot and share it with your users, you don’t simply let it fly alone forever. Much like a mother cares for its baby, so too must you nurture your chat baby and help it grow to its potential. When a bot is born, it is assigned a bot trainer to manage the day-to-day interactions that take place. In the analytics dashboard of your platform, you can read up on the transcripts your chatbot has had with its users over the course of anytime. Especially in the beginning, the chatbot is going to receive many user inputs it doesn’t understand. These unhandled messages get fed to the bot trainer to correct the chatbot. It’s a learning model in the AI space we call supervised learning. As time passes, your chatbot becomes smarter and needs less hand holding to do its job effectively. Monitoring your chatbot performance helps you train it to be better at serving your target users. Data = Power!
Chat is changingSmart people do research before venturing into the unknown. They grab a roadmap and plot a course to succeed in their goal as best they can. To summarize the things you should keep in mind when getting your first chatbot up and running:
- Explore chat use cases beyond just customer service. Probe your departmental teams for conversational pain points. You might discover there’s a need for more than just one chatbot to be built. If that’s the case, prioritize which use case can bring the most value.
- Find a platform to build your chatbot off and invest your time in the user experience. The less time you think about programming (including integrations), the more time you can focus on the most important reason for building a chatbot: the end-user.
- Dedicate a person to training your chatbot. It’s not a one-and-done exercise and needs frequent check-ins to optimize performance. Remember the trade-off here is that a single chatbot is servicing thousands of end-users all at once.