As products seek more ways to engage their consumers, product designers are clamoring to create chatbots. Perhaps you are one of them, and are considering whether to use a chatbot as part of your business model. You may have even taken the next step, and are researching vendors to create one for you, or crafting your own using open source tools.
Regardless of where you are in your chatbot journey, it’s important to understand the capabilities and the limitations of chatbots – they are not a one size fits all solution.
If you are an established brand with decades of market awareness behind you, then it may actually be more difficult for you to convince or train your customers to use a chatbot for their self-supportable service needs. As they are already familiar with how to use and engage with your product, introducing this new feature can have low adoption rates.
But if you are a small – perhaps not yet well-known – brand, you have more flexibility in setting expectations with your customers from the beginning.
Chatbots: What to Know and Where to Begin
Chatbots are not people. Chatbots and the technology underlying them are currently quite adept at a few types of simple conversations – questions regarding password resets and other “simple” questions which are frequently asked.
Generally it’s a good idea to start with this type of conversation or situation which lends itself to the technology. Trying to push the technology to do things it’s not extremely adept at doing yet can be disastrous.
Here are some steps to take when designing your chatbot.
1. Figure out the conversation.
Start with one scenario. First ask yourself “what.” What is the conversation about and what is the answer?
2. Understand your customer.
Research, explore, and test how your customers go about trying to solve their problem, and identify what words they use in that “help me!” conversation.
3. Test it for normal use cases.
Test your chatbot for tone and language- does it sound like a robot or like it could be an actual person? You want to aim for the latter.
Is it able to answer the type of question/scenario you’ve designed?
4. Test it for bizarre use cases.
Remember that your customers are humans- and humans will be human. Movies like Her illustrate how our expectations of chatbots, robots, and other conversational interfaces are still being tested. We are so delighted with the mere technology existing, that – not unlike kicking the tires on a new car- people are actively trying to see what these experiences are capable of. What happens if I swear at it? What if I ask about its favorite color? Can I fall in love with it? This may seem odd, but I guarantee you that everyone you are sitting with right now- including yourself- has asked Siri some seriously weird questions.
Thus, before launching any conversational UI, it is vital that you have tested out not just the conversations that you know it should be able to have, but have also found an elegant way to handle questions about absolutely everything else.
Handling questions doesn’t mean your chatbot needs to have all of the answers – it just should be able to gently guide the customer back to what your conversational UI can talk about. This is done with the use of what’s called conversational guardrails. These guardrails might be verbal/text only, or may include visual “cards,” showing someone what topics it is capable of helping them to explore.
These are just a few of the very high-level basics behind when and how to create conversational UIs for your business. If you want to dive in deeper, be sure to check out my webinar, click here!