...

How to Make Rule-Based Bots Sound More Natural?

Illustration: Jana Pérez
rule-based bots conversation design

Please note that 'Variables' are now called 'Fields' in Landbot's platform.

Please note that 'Variables' are now called 'Fields' in Landbot's platform.

There’s a lot of talk on the web about designing conversations for AI chatbots. After all, designing a sensible narrative structure for AI is quite the feat. However, in all that upheaval, conversation design for the simpler rule-based bots has been forgotten.

You must be thinking that designing conversations for rule-based bots can’t be that hard… Well, that indeed should be the case. 

Yet, here we are! 

In reality, a lot of rule-based bots fail to reach higher than the chat format and often get stuck exhibiting the same unfeeling and rigid approach of their “second cousins” AKA online forms. 

Being tired of seeing bots spitting out copy-pasted text from forms and landing pages, we decided to put together this conversation design guide specifically dedicated to rule-based chatbots. After all, they deserve our love too.


1. Ensure Turn-Taking

One of the core pillars of a functional conversation is so-called turn-taking. While it might seem quite obvious that conversion is not a monologue, many forget to give users space to actually interact when designing a rule-based bot. 

Have you personally experienced a bot that bombards you, after just one question, with a tedious “rant” that just goes on and on?  A monologue providing little diversification from actually reading a landing page or a list of FAQs? If not, well, it’s not fun. The whole conversational principle just goes out of the window. 

So, if this is your first rule-based bot, don’t forget to actually have a conversation with the user.

Large chunks of text such as these inside the bot are not fun (you might as well stick to normal landing page):

In your defense, you might argue that the answer you need to give simply is long and cannot be simplified or cut in half. In that case, design for a respite for the chatbot user… in the middle of your explanation let the bot stop and ask “There’s more but how are you doing so far?” “All clear so far? “Are you ready to hear more?” etc.

Here is the same text but now interlaced with the user interaction triggers that keep them engaged:

2. Don’t Teach Commands or Give Instructions 

Because rule-based interfaces often rely on rich user interface elements such as buttons, carousels, enriched fields, or images, it tempts chatbot makers to stick to the usual form-like commanding language. 

However, conversational interfaces are so impactful mainly because they feel natural. If you strip your chatbot of that, you are missing out!

In other words, if you need to “teach” or “tell” people how to use it or answer your bot, you are doing it wrong. If it’s not natural, it’s not working!

‍For example, this is a teaching order you should avoid

“Select one of the options below or click to see more options.

“If you want to see other options type/click ‘Other Options’?”            

Nobody talks like that. So, instead, go for something more natural such as: 

“Would you like to see more options?”


3. Avoid “Software Statements”

Software statements are statements that pop up from software or digital tools such as your email interface. 

To stay true to being conversational, you should stay away from impersonal statements such as “The email you entered is invalid.”
Yes, it conveys the meaning, but you are supposed to be having a conversation so… soften it up with something like: “Hmmm, there seems to be an issue with the email address you gave me. Try retyping to make sure I have it right.”

The same goes for the “Button” language. Just because you are using buttons, you don’t have to resort to button language like “Confirm” “Send” “Accept”.

Instead, design buttons that fit your audience:

“Yeah, of course!” 

“I totally agree!” 

“Looks OK to me 👍”

“No, it’s not necessary”

“Please show me the options again.”


4. Use Discourse Markers

You don’t need to be a copywriter to make your bot copy sound a bit less strained and conversational. All you gotta do is pepper around a few discourse markers.

Discourse markers linguistically or emotionally relate the upcoming phrase to what was previously said. They help comprehension and make the conversation more fluid and less robotic.

Common examples of discourse markers include: 

Also - Actually - Ironically - Above all - (un)Fortunately - As an example - In that case - In other words - By the way - Consequently - As a matter of fact - For that reason…

Examples:

  • By the way, did you know that this week only we offer 30% off for all new customers?
  • In that case, let me share the booking calendar so you can schedule a meeting!

5. Use Emoji and Other Rich Content (Sensibly)

The world of instant chat is filled with fun possibilities that take on the shape of emojis, Gifs, icons, images, carousels, and videos into the conversation.

Emojis and other rich media formats allow you to make up for the missing emotions, gestures, and expressions that make up face-to-face conversations. They can revive any chat, even the rule-based kind. 

Be creative.

You can even leverage rich media to substitute the monologues and the not-so-conversational long chunks of text I mentioned above. 

Images and videos often speak for themselves.

However, don’t overdo it! After all, it’s still supposed to be a conversation.

Also, always remember the context. For example, ask yourself: “Are your users usually in an environment that allows them to watch a video?”


6. Read the Conversation Thread Out Loud

It might seem silly and too simple to be of any help.

Nonetheless, it’s a crucial step in developing a fluid conversational experience.

After you are done writing the conversation thread, read it aloud and, if you can, get a second and even third opinion on it.

Better yet, you can ask some of your best customers to test it for you.