Doing something for the first time can be scary. I remember when I was little my mom bought me a bicycle in our little suburb home. Up to that point, I was comfortable relying on my two legs to get me where I needed to be. But all my friends were moving around on these 2-wheeled vehicles, getting to places much faster. What would take 30 minutes to get to the park could be achieved in just 5 minutes with a bicycle. So, I took the plunge and learned a new skill. For many of us, adopting new technologies into our business operations is essential to growth and efficiency, especially with limited resources at our disposal. Do more with less, as the common wisdom coined by R. Buckminster Fuller says. Ephemeralization is a key to meeting the ever-increasing standards of living and the ever-growing population of your customers, employees, and stakeholders in your company. Acknowledging value in tech adoption, the challenge now becomes how you adopt it in the workplace as a standard practice. Depending on the tech, the learning curve may be steeper than others. You’ve expressed interest in chatbot technology, and now you get to learn the necessary steps to implement your own AI conversations with your audience by following this step-by-step guide.
The StepsIn the following steps, start by crafting your chatbot documentation into a Bible of sorts, which will serve as a reference for you to fall back upon and iterate over the chatbot lifetime. I call it The Bluefish ICONIC Approach.
#1 – Identifying Problems and GoalsProblem Statement: Demonstrate the user experience for a chatbot to provide to your core audience. User Experience (UX) begins with the end-user in mind. The first step in crafting meaningful chatbot conversations is to talk with people who are having communication issues with the company and identifying that a chatbot solution will solve those. Borrowing from the old journalist trade book on hard news, ask the 5Ws and 1H.
- Who is the user?
- What is the goal/motivation behind the user interacting with the chatbot?
- When will the chatbot be used?
- Where will the chatbot be used?
- Why should they use the chatbot over other tools?
- How will the user interact with the chatbot?
DeLoyn Luxury Watches
- Our users are customers who shop online or in-store.
- The user’s goal is to gather additional information about DeLoyn and its products in order to make an informed purchase.
- The chatbot will be used whenever the customer visits the company website, social media channels, or on the in-store kiosk. It will be available for service any time of day or night but will offer to connect with a human customer service rep on the company website between 9am-5pm on Mondays through Fridays.
- The user will initiate a chat by clicking on the message bubble, read through bubble descriptions, click the bubbles that are of interest until the user reaches his/her goal.
- The user will use the chatbot over other methods of communication because its convenient, responsive, and easy to use.
#2 – Capturing User StoriesIf you’ve ever watched television, there’s a break in between the programming with advertisements. Some of the ads may resonate with you while others are completely irrelevant. Traditional selling with a shotgun approach has changed dramatically since the 20th century and it’s no longer acceptable to bombard people with useless material they don’t want. When crafting your chatbot, you should always make the conversation very specific to the intended user you want to engage with. If left to its own devices, your audience will surely disconnect, frustrated, and upset from the whole experience. So, take the time to segment your audience into different user stories and craft multiple chatbot scenarios that meet those goals you established earlier. Consider the “DeLoyn Luxury Watches” example. The company may decide to expand chatbot offerings to employees, who might have their own personalization traits. It might make more sense to create a separate chatbot that is employee-specific, and one that is customer-facing—each with their own scripts. What might be likely is the segmentation of services needed. One customer might interact with the chatbot to make a direct e-commerce purchase, while another customer might want to complain about a defective watch. Contrary to popular belief, having a catch-all solution rarely works and the best tools solve specific problems for people. This is why segmenting not only your users but matching them with their respective chatbots will be an immense help in providing a meaningful and effective experience that brands your business as trusted and helpful to its users.
#3 – Organizational Landscape UnderstandingStep 3 gets into the nuts and bolts of your technology stack by answering the question of what you have and what you need. If you are fortunate to have your own engineering team, you may be able to create a chatbot from scratch. But many businesses rely on vendors to build chatbots for them. There are typically 3 ways to go about it.
- DIY – build from scratch
- Platform – drag & drop
- Templates – copy/paste
- If you are putting your chatbot on your website, who are you hosting with?
- If you are integrating with any 3rd party programs (CRM, ATS, Calendars, etc.), do they have open APIs to connect with?
- If you have security concerns, what kind of protections do you need in place?
#4 – Natural Language Conversation DesignNow for the exciting part. You can create the tangible conversation experience you want your user to have with your chatbot. There are 2 major elements to this step—the UI and the chat itself.
The user interfaceThe visual elements that make up a chatbot are just as important as the text itself. UX designers create wireframes to visualize the experience. You can start with a low fidelity wireframe that addresses the information hierarchy insights from your previous research. For this exercise, I highly recommend using the classic pen and paper to sketch out what your ideal design might look like. For those wanting a digital UX tool, you can try out popular ones like Figma, Adobe XD, and Sketch. Visual Hierarchy as a Framework for Chatbot Conversations Three common elements of chatbot UIs are the header, body, and footer.
- Header – Here you establish the bot identity. This might be promoting the company brand, the purpose of the bot, or the logo of your business to reinforce your position in the market.
- Body – The bulk of the real estate is here. You establish the 2-way conversation between user and bot, timestamping a reference, and recording to review post conversation.
- Footer – Here is where user inputs can be added. This could include free text or button commands.
The conversationHow would you want to be greeted when you log onto a company website for the first time? Say you call the customer hotline to DeLoyn Luxury Watches. There are certainly do’s and don’ts of proper etiquette that affect your experience. Here are a few things to include:
- Salutation/Greeting – How does that first message make you feel? Welcomed? Intrigued?
- Feedback Loops – Instead of hitting a dead end, how might you keep the conversation going once it reaches the end of a goal?
- Simplicity – Too much jargon can bog down the real estate of the page. Use easy to understand words.
- Bot/User Icons – Puts a face to a name and makes it more relatable. Notice in the example, we gave our bot a name “Lindsey”. While it’s not a hard-fast rule, typically adding a personality to your bot that aligns with your brand will create for a more engaging experience.
- Line Breaks – Long paragraphs of text can be hard to read and may make a user disconnect from the conversation. Keep it simple!
#5 – Implementation of ChatbotUp until this point you’ve been blocking out the scenes and rehearsing the script. Now it’s showtime! If you used the platform option as I mentioned, there are a couple of areas you will need to go to implement your chatbot. In this step, I’ll be using the Bluefish.AI platform to demonstrate actionable areas for implementation, but many of these steps can be applied on other platforms too. The Brand Themes Screenshot of Bluefish.ai console. The FAQs FAQs Screenshot of Bluefish.AI console The NLP NLP Screenshot of Bluefish.AI console The Integrations Integrations Screenshot of Bluefish.AI console
#6 – Continuous ImprovementCongrats! You’ve successfully created your chatbot and now users can engage for a satisfying experience that will catapult your business into a more productive position. But there’s one final step that’s often overlooked. Chatbots and the conversations they have with their users are learning experiences that give you invaluable insight into the hearts and minds of your audience. You can learn all manner of things like buying habits, pain points, and more. So as your final task, you must constantly monitor the chats and iterate on them. Here are two fine examples of what I mean for chat improvement:
- A user is asking the chatbot a question and he/she keeps getting error responses that the chatbot doesn’t know. This is a golden opportunity for you to “train” the bot on the correct answer to reply with future interactions on that particular situation. One of the great myths of AI is that they are omnipotent and this is simply not the case.
- Job applicants keep asking about compensation regarding the position. While your chatbot answers generic inquiries on salary, you see an opportunity to expand a whole section of the user experience. Touch base with the hiring managers to address a holistic answer to one area of the candidate user story. Details might include vacation, retirement, discounts, etc.