Due to the variety of types of chatbots, no article will be comprehensive enough to fully explain how every chatbot operates. We will explain how chatbots work on the most basic level.

Trigger/Calling: Most chatbots will begin by being called or triggered. For example, if you want to speak to begin an interaction with the Amazon Echo, you would simply call the name, "Alexa," and say your command. If you want to interact with a Facebook chatbot, you have to go to Facebook, tap on Messenger, then use the search bar to find them. Once you open it up, a Facebook bot chatbot will begin talking to you.

With our platform, you can call a chatbot in a variety of ways--this could be by tapping a specific button, or through a behavior such as staying on a website page for 35 seconds and scrolling twice. Simply stated, you must initiate a chatbot and let it know you are ready to chat!

"Ready" Confirmation: Once triggered or called, chatbots will often greet you and let you know that they are ready to get started. They may say, "Hi", or ask you what they should do next, such as "Would you like to purchase lunch?'".

Command/Input: When your chatbot has confirmed it's ready, a user of a chatbot will say or write a command. For example, it may ask if you like a bean burrito, or a taco supreme, and wait for you to tell it or type a response.

Command/Input Confirmation: A chatbot will likely confirm that they understand the response, or have the correct response. In an audio-based chatbot they will often repeat the command back to you. In a text-based chatbot, they will often record the response so there is a clear and written record of the conversation. (In our chatbot, all responses are recorded in a user's profile, so you can easily review how people have responded to chatbot conversations over time.)

Logic/Processing: Once the response is received and confirmed, the chatbot will process both the intent of the response (what is the command, what is the need), and also comprehend what to do next. There are several systems which can be utilized for this part of the process, mainly rule-based logic, machine learning, and/or artificial intelligence--explained above in What is a Chatbot.

Chatbot Response/Action: Once comprehended and "thought-about", the chatbot will perform a command, action, or goal (e.g. purchase a lunch).

This process can be very dynamic based on user responses and actions, as well as behaviors and/or perceived inputs (e.g. temperature outside, sound of dog barking, etc.)

Your chatbot strategy will depend on your company goals. The best thing to do is to ask yourself some basic questions:

• What is the ultimate goal of your chatbot? (distribute content, create purchases, explain directions, customer support, etc.)
• Who will be creating and managing your chatbot? How much time will they have to do so?
• What analytics are you looking to get from your chatbot? And where do you need this information stored?
• How consistently accurate do you need this information to be? (e.g. if there is an NLP error, will your users be flexible with this error?)
• What platform will best serve your users? (e.g. Is it for internal purposes? If so, do your employees often use Slack?)
• How much time can you devote to managing your chatbot? How big of a priority are chatbots for your company?
• Will your require images or videos in your chatbot?
• Do you need your chatbot to upload and save documents?
• Will you require your chatbot to have REST API capability so it can "talk" to your internal systems (e.g. inventory or CRM systems)
• Do you have a separate marketing budget to launch and acquire users for your chatbot if it's on external platforms (e.g. Kik, Facebook, etc.)
• How will your users utilize your chatbot? Will they request simple direct tasks, or might they be varied and nuanced? (This will help you factor in where and when you leverage natural language processing).
• What will be the most convenient way to use your chatbot? Through voice or audio or both? And if both, at what point will you tell the user to switch?
• Who do you need to have access to the data of your chatbot? Will it just be within the tech team's domain, or will the marketing team need full access to both editing and analytics?

Answering these questions can help you decide if you build a custom chatbot, or a use an existing platform. It will also help you decide which platform is best and why. We're happy to talk through some options. Even if we can't help you, we have many partners who may be a good fit. Email us at [email protected] with the subject, "Chatbot Strategy".

Sign Up Now