July 9, 2017

Conversational vs Button Based Chatbots


When implementing a Chatbot for your company there are two major directions you can go. First is having a button based chatbot and the second is implementing a conversational-based chatbot. Which one should you choose? What are the pros and cons of each one of them? We at Marvin will try to give you a short overview of the two main parts that small and medium businesses should pay attention to – costs and usability of the chatbot.

Let us have a very simple case with the following assumptions:

  • You are trying to build a chatbot for a travel agency (a travel assistant chatbot) that will tell the user what are the visa requirements and potential travel warnings.
  • The user wants to travel to the United Kingdom.
  • We assume we find the users country of residence via IP.
  • The user types in English.


Costs for Chatbots:

Chatbot costsChatbot cost 


Button-based Chatbots  are generally easier to program and you will not need to break the bank to get one. Also, the implementation time is a lot shorter, both in terms of programming and in terms of data, since you guide the user through the experience by allowing him only a certain set of answers. For our case, all you need to do is offer the user the possibility of choosing between predefined lists of countries


Conversational Chatbots



Usability of Chatbots:

Chatbot usability

Chatbot usability

If a button based chatbot has usually more than three buttons, then the whole process of using one becomes annoying. In such cases, users prefer using a website rather than scrolling up and down in a chat interface. To put this situation in our framework, there are currently 196 countries in the world; can you imagine your user scrolling through them all? Obviously, you can apply certain filters or limit the countries. Still, the whole procedure is annoying to the user.

Conversational Chatbots obviously have an advantage in usability as opposed to button-based Chatbots. If you follow best practices you can predict most of the user inputs.  Further, with proper monitoring, you can also generate further logical connections between user input and proper answers, thus greatly increasing the accurate response rates. To look at our case – let’s say you have controlled all options but “UK”. After some interactions, you become aware that “UK” is also a term that should trigger your chatbot to provide visa requirements and potential travel warnings for the United Kingdom, and you integrate it in the logical connectors.

What is the conclusion from the notes above? We leave this question open-ended for readers of our blog to decide what the best way is. A button based chatbot? A conversational one? Or maybe a mix of two?



Leave a Reply