You can think of a flow chatbot like a yes/no conversation. It goes like this.
1. The user triggers the chatbot with a pre-set keyword (i.e. 'Start')
2. The chatbot responds with the autoresponder, asking the user to choose the next step.
3. The user picks the next step, usually by clicking a button or sending another keyword.
4. The chatbot responds to that specific keyword with the pre-set answer
A flow chatbot essentially never ends. You could continuously go around the conversation by resending the start keyword and pressing the buttons.
Finally, a flow chatbot does not use any artificial intelligence. It doesn't learn or understand human text/voice input unless the owner modifies the code.
A simple chatbot flowchart
Let's try to put all of this into a nice flowchart. Hopefully, this should help you visualise the messaging sequence and the logic behind this very simple type of chatbot.
Disregarding my inability to draw pretty flowcharts, this is what happens.
In the centre of the image, we have a user input. This is when the user sends a keyword that triggers the chatbot.
As an example, we built a flow chatbot as a prototype and proof of concept for one of our clients that had the keyword 'SEO' as a trigger. Sending 'SEO' would trigger the first message from the chatbot which turns into a full on SEO course delivered via Facebook Messenger.
So, the user sends a message.
If the message is unrecognisable by the chatbot, we follow the left-side branch and hit the auto-responder / human fallback.
Typically, all the chatbots we build have autoresponders, fallbacks, and eventually a human fallback. An autoresponder is a message that simply welcomes the new users and shows a series of keywords or commands they can use.
A fallback is usually triggered after a user has sent several unrecognisable inputs in a row, again educating the users on the commands they can use.
Finally, a human fallback is exactly what it sounds like. After multiple unrecognisable inputs from a user, the chatbot sends a notification to a human to nudge them to take over.
If the message is a recognisable keyword, the flow follows the right-side branch. This is where you can see the decision-driven interaction between the bot and its user. The bot presents options, the user picks one, then bot then presents the next set of options, and so on.
Ultimately, either the user has reached its destination and found what he was looking for, or the bot's fallbacks have been triggered.
There we have it, a flowchart representation of the simplest form of chatbot!