4 Chatbot Building Secrets from a Conversational Design Expert
A psychologist, a computer scientist, and a novelist walk into a bar…
Don’t worry, this isn’t the start of a bad joke. But it IS the beginning of your chatbot dream team!
Why do you need three people to make one bot? Well, it isn’t so much about the people as it is the skills these people embody. In a recent podcast with Hans Van Dam, the co-founder of Robocopy and the Conversational Academy, Hans shed some light on the chatbot space and why it should be taken seriously.
“There’s a right way of doing [chatbot design] and a wrong way of doing it. I think that’s what we need to educate the market on. It’s a serious job and it requires serious training.”
Chatbots are relatively new to the business space, which means that there aren’t any precedents for how to create and manage teams responsible for the design, implantation, and upkeep of chatbots. The technology is out there, but it’s being underutilized because no one really knows what the best practices are. One great place to start? Creating multi-dimensional teams that hit all aspects of a successful chatbot.
The Dream Team
Let’s go back to the psychologist, computer scientist, and novelist. One understands how humans work, the other understands how the technology works, and the final one knows how to put it all into words. All of these elements exist within a chatbot. At its core, a chatbot’s purpose is to engage a potential customer in conversation. When humans have conversations with each other, we utilize innate linguistic and psychological signals to communicate with and understand one another. These elements still need to be present within a chatbot, in addition to the nuts and bolts of actually getting the chatbot to work.
Finding all these skills within one person is rare, which means that for larger companies, it’s essential to create diverse chatbot teams that include talented people from each field. For smaller companies where it might not be feasible to hire a huge team dedicated to chatbots, try to find someone who is strong in one or two of the mentioned areas and seems capable of learning quickly, and send them to trainings (like the Conversational Academy) to round out their capabilities.
So now that you have your dream team all trained and ready to go, it’s time to get to the actual designing and building! The goal is to create a bot personality that’s consistent and trustworthy. Think of the interaction you’re designing as a true conversation: both the human AND the bot need to get something from the conversation. Humans have rational and emotional needs: we generally have certain motivations, anxieties, and expectations that we want met. Bots, on the other hand, are a little more straightforward. They usually need to relay certain pieces of information and ask specific questions, hopefully collecting the information associated with those questions. Chatbots should be built to proactively anticipate the user’s anxieties and motivations so they can calm the anxieties and reinforce the motivations. This way, the user’s needs are met and the bot gets the interaction and information it needs, making the conversation productive for both parties.
Steps for Success
Now, the theory of bot design and construction is great, but what does it look like in practical application? Hans Van Dam shared a few tips that will help you refine and improve your chatbot design process.
1. Create an empathetic bot
Write your bot to appeal to the humanity of the user. Establish empathy early on in the conversation by using what you know about your users to address their potential boosters, barriers, and information needs. When the user feels understood, they’ll be a lot more comfortable letting the bot steer the conversation. When there’s no empathy, the other party starts talking a lot and that’s when the chatbot breaks because it’s not equipped to handle that kind of interaction.
Once you establish empathy you can have a very simple dialogue with short and direct questions. Empathetic designs tend to be longer than other more typical designs, but they work better because they develop trust between the user and the bot.
2. Make your words count
Try using what Hans calls the “one-breath” test. If it takes more than one breath to read one turn of the chatbot, that message is too long. Shorten your messages using Hans’s “Jenga technique.” Jenga is a game where a tower is built with stacked blocks. The object is to remove as many blocks as possible without collapsing the tower. Your chatbot messages are essentially a tower of meaning. Once you’ve written a message, start taking out as many words as possible without distorting the meaning. You’ll find that most of the time you can reduce the words in a message by up to 50%.
3. Don’t forget to incorporate personality and psychology
Once you have your core message, you can spruce it up with different design techniques. Make sure there’s a confirmation or affirmation in every speaking turn the bot takes and end each turn with a prompt so the user knows it’s their turn to respond. Make sure your represent your brand through the vocabulary you use. And don’t forget to mange the user’s expectations and add social proof to encourage engagement and completion.
For instance, instead of immediately asking the user questions once they’ve signaled they want a demo, try saying something like, “Great! I just need to ask you three quick questions. For most people, this takes less than a minute!” This statement lets the user know they’ll only have to answer three questions and it gives them proof that it hasn’t taken other people very long to complete the questions.
4. Make sure your conversations sound natural
When you’ve fully designed a conversation, test it out to see if it sounds natural! You could try doing something called the “Wizard of Oz” test. Write the conversation out on sticky notes and put them on a wall. Then bring in five to ten people who aren’t involved with designing the chatbot and have them respond to the conversation. Let them say whatever they want. Bonus points if you put the two people having the conversation back to back so they can’t rely on body language to understand the context! This exercise will show you pretty quickly what is and isn’t working with your bot because you get to see exactly how a stranger will interact with only your bot’s words.
Bot building is a real skill that deserves to be taken seriously. When implemented correctly, it can be a powerful tool for your business. It will take time, so be patient! You weren’t an amazing driver the second you earned your license. Just having the right software won’t make you an expert on bot building, but it will put you on the path to becoming one.