With the introduction of chatbots, a new way of interaction has appeared: we can use them in a more natural way: by writing messages.
This gives users a greater degree of freedom, but this freedom can easily kick back.
The chatbot technology is developing rapidly, but it is, by far, not limitless. We have to keep that in mind, but not only us, developers; so do our users.
Let’s see what are the Dos and Don’ts if you want to reach the best chatbot user experience!
#1 — Draw The Lines Of Usability
Many may feel enthusiastic to be able to express their problem in words and the bot responds to them.
There is nothing wrong with that since that is the very essence of chatbots.
However, we cannot expect a user to know the operational background of a chatbot, nor do they have to know where the limit is while the bot can understand the problem.
Although it is quite understandable and natural for someone to make a very long and detailed description of their problem with the opportunity of verbalism.
A person could easily understand those descriptions, but not a chatbot. The technology is simply not quite there yet for a machine to understand text with complex logic connections.
It is our responsibility to handle this problem. We need to teach our bot how to react in a situation that they cannot solve on their own.
But even better if we try to avoid similar situations and design an UserFlow that clarifies how to use the chatbot.
# 2 — Make Clear What The Chatbot Is Good For
Let’s be preventive: at the very beginning of the conversation, introduce users to what and how they can do with the chatbot.
Here are two solutions we used for chatbots we have developed:
In Sihuhu train timetable chatbot, we briefly summarize in the first message what message we expect from the user to make the bot work properly.
The bot then offers the option to provide a more detailed description of functions, and the user can decide if they need it or not.
AcerShop.hu has a bot with complex and versatile functionality. We present available features as cards, so users can scroll through them. They can then, if necessary, request instructions for using each function separately.
#3 — Do Not Pretend To Be A Human
It’s crucial to clarify at the beginning of the conversation that the user is not talking to a living person.
If you get an overly realistic character, the chatbot may be misleading and users will be more likely to describe a complicated issue that the bot will not understand.
This will result in a negative experience at the very beginning of the conversation.
#4 — Tell What The Problem Is
The ability to learn from our own fault is a wonderful skill of mankind.
To do this, however, it is essential to know what the error was.
For example, if a user asks a question that is too complex, the bot, instead of responding incorrectly, can suggest them to make it shorter and simpler. Next time, users will know how to communicate with a chatbot.
#5—Learn From The Users
In addition to learning from our own mistakes, we should analyze the behaviour and expectations of our users.
It is important to check from time to time how people use our chatbot. We can draw lots of lessons so we can continuously improve our chatbot. For more about that, read our story about maintaining your chatbot.
Disclaimer: this is a translated version of an article originally posted on our own botblog.hu, on May 29, 2019.