Maybe you’re looking to deploy a chatbot. Maybe you’re interested in learning more about the technology. In either case, you’ll likely come across the term ‘flow-based chatbot’ from time to time.
There are different ways to get chatbots to communicate with users. Flow-based is one of the simpler ways to get bots talking.
But with AI chatbots on the rise, is there much reason to opt for flow-based bots? Here, we explore the pros and cons behind using a flow-based chatbot.
What is a flow-based chatbot?
A flow-based chatbot is one that works by using a pre-defined (and reined in) conversational flow.
In other words, they have their conversations already mapped out like a flowchart. When a customer triggers a conversation, the chatbot guides them through the conversation flow chart, step by step.
As an illustration, a flow-based chatbot of an online fashion retailer might cycle through closed questions and subsequent suggestions like:
- • Occasion (“What’s the occasion you’re shopping for?”)
- • Clothing type (“What kind of item are you looking for?”)
- • Size (“Okay, what’s your size?”)
- • Showcasing samples (“Okay, how about these items?”)
For the user, the bot would typically offer available answers as buttons. So, in the fashion retailer example, the user would just click a “Wedding” button, or a “Coats” button, for instance. Or, for limited queries (i.e. age or gender), they can rely on the user typing in specific keywords.
Either way, the flow-based chatbot takes a systematic approach to a narrow conversational field. Rather than type and go, the user is guided down a set chat flow.
Pro: Clear abilities
A flow-based chatbot’s conversations work toward pre-defined goals, which the bot can tell the user at the start of the chat. This means that from the offset, it’s clear to the user what they can achieve by interacting with the bot.
Plus, the bot guides the user through the conversation, supplying viable answers and asking closed questions. This means that it’s easy to help the user avoid running into a brick wall. If the user doesn’t offer a reply the bot can understand, it can tell them the answers it can reply to.
Con: Limited options
Although the abilities of the chatbot are clear, they’re also restrictive. Users cannot diverge from the intended conversation options at all. The only way for a flow-based chatbot to hold a conversation is if you’ve provided it with the exact script and decision tree.
This means that a user may find themselves disappointed if the chatbot isn’t equipped to handle their query.
Pro: Maintain context
It’s easy for a flow-based chatbot to appear to remember the context of a conversation. Remember: they follow a scripted decision tree. So, their pre-written replies can reflect the previous answers in the conversational flow.
This saves the user the frustration of dealing with a confused bot that keeps forgetting what they’ve told it.
Con: Unnatural conversation
A flow-based chatbot works using specific keywords and prescriptive buttons. In situations where a user types a reply to the bot, if they haven’t used any of the exact keywords, the bot won’t understand. It cannot derive meaning from natural language.
This means that it can become difficult for users to give the ‘right’ answers that will further the conversation.
You design the conversation and write all the replies when deploying a flow-based chatbot. So, you know that if the user chooses option X, the bot will always respond with reply Y.
For this reason, it’s much easier for you to guarantee the experience your chatbot will offer. It’s always going to follow the same conversational flows.
Con: It can be boring
Flow-based chatbots can’t adapt or change their messages based on extra information. (I.e. the tone of the message or extra details the user provides.) They never stray from their defined path.
This can make for not only a potentially unempathetic response from time to time, but also a boring experience for any regular users.
Pro: Easy to tweak
If something isn’t working the way you want it to with a flow-based chatbot, you can easily tweak, edit and add to its decision tree to fix the problem.
For example, you can add more keywords to help the bot become less likely to get stuck. You could even add a whole new conversational branch for the bot to offer users. There’s no black box problem to deal with, as you might find with a machine learning chatbot.
Part of a whole
A flow-based chatbot comes with limitations. It’s relatively inflexible in its offering, but that inflexibility can provide other benefits. It allows the bot to offer predictability and keep conversational context.
With the rise of AI chatbots, there’s still a place for their flow-based friends. In fact, as chatbots continue to evolve, AI will likely bring adaptability and natural language to increasingly powerful flow-based bots.