Conversation has become the new “king” of online marketing. Customers no longer want to passively consume polished advertising claims. They want to take part, they crave to experience what your brand is about. Moreover, they want to feel an emotional connection that will solidify the “correctness” of their choice. In other words, the experience economy trend has changed the marketing landscape and brought us to the foothills of conversational design.
That’s why you are here.
The talk of and interest in conversational UI design is not entirely new. However, with the increasing ease with which we can create conversational experiences has opened this topic to a much wider audience.
While we have become masters of online content, subduing the arts of SEO, readability and user-friendly formatting, creating conversations has left many business and professional writers at a loss. For one thing, conversations don’t seem to have rules. They are unpredictable, more personal and the use of colloquial language often goes against instincts when trying to create an image of authority and expertise.
So, we put together The Ultimate Guide to Conversational Design to provide a holistic resource for those interested in tackling the art of conversation and understanding the role of language in design.
The guide will:
- Explore the reasons for the sudden rise of conversational interfaces (AKA chatbots);
- Define what is conversational design;
- Identify its three pillars;
- Outline the fundamental elements of conversational UI;
- Dig into the crucial steps of conversational UI creation prep;
- List the Dos and Don’ts of writing a bot conversation that engages and converts
Before we started using chats and messages to talk to bots, we used to talk to each other.
The stats clearly show that our society has become strangely fond of texting, messaging, chatting – whatever you wish to call it. In the past decade, the number of monthly sent and received texts sent has increased by over 7.700% in the US. In fact, 97% of Americans text minimum once a day.
Texting can be truly immersive. In fact, we are so compelled to attend to our messages that we often do so at the risk of our own health and safety while crossing a street or driving. In this, we are all guilty. According to a study by Twilio:
- incredible 82% of consumers keep messaging notifications ON;
- An average person has 3 messaging apps on their smartphone home screen and uses 3 different messaging apps per week.
Messaging, though completely technology-enabled has become a fundamental part of human experience.
Erika Hall, in her book Conversational Design, argues that the attraction of texting has little to do with high-production values, rich media, or the complexity of the messaging features. Instead, she claims, it’s the always-accessible social connection, the brevity and unpredictability of chat conversation that triggers the release of dopamine and motivates to come back for more.
Consumer fascination with chatting led businesses and marketers to believe that conversation is the best model to engage and captivate our ever-more fleeting attention.
They were not wrong.
Coupled with hunger for more personalized experiences and emotional connection with brands brought about by the experience economy, conversational approach has started to win over consumers all over the world.
Conversational design evolved from simple conversational CTAs such as Facebook’s prompt to post an update:
To fully interactive conversational user interfaces:
Since conversation is intrinsic to our daily existence, the more an interface leverages its functionalities, the less you need to teach your visitors how to use it.
However, Hall further elaborates that while the experience starts on screen, the real magic happens in our minds. We consume these brief messages riddled with subtle linguistic hints and our mind translates them into personality, humor and coherent narrative.
This is why trying to be conversational intentionally is not that easy.
Conversations are immediate and painstakingly dependent on context. Hence, artificially creating a natural sounding flow takes more insight than it’s apparent at first glance.
That’s why it’s important to regard conversational design as its own discipline.
You can decide to adjust your website’s copy to leverage conversational principles like in the example with FB post prompt. Or, you can build a whole CI. Either way, it’s important to understand that conversation design is not a simple act of writing down text in a conversational format.
Similarly to the process of designing a website or writing a book or a movie script, it requires a complex set of skills and careful planning. Conversational UI design is, in fact, a combination of several disciplines including copywriting, UX design, interaction design, visual design, motion design and, if relevant, voice and audio design.
Google compares the job of a conversational designer to that of an architect who maps out what users can do and achieve in a particular space while considering user’s experience, needs and technological limitations. In other words, conversation design not only requires the use of natural conversational language but also creating logically sound conversational flow and its design specifications that capture the entire user experience.
Before you go off designing chatbots, you need to get comfortable using language in conversational context The first step towards that is understanding the three core pillars that make conversation a conversation:
I. Cooperative Principle
The cooperative principle was first phrased by philosopher Paul Grice in 1975 as part of his pragmatic theory. According to this principle, effective communication among two or more people relies on the premise that there is underlying cooperation between the participants.
In other words, instinctively, we don’t just act cooperatively ourselves but we also expect this kind of behavior from others.
Let’s break it down.
Take a look at the example below:
Alice: “Do you know who is invited to the wedding?”
What do you think about this conversation?
It’s completely grammatically correct.
Yet, it’s weird. It’s rude.
That’s because Peter is ignoring the cooperative principle. Yes, Alice did ask a simple yes-no question. However, the question implies she is expecting Peter to tell her who is invited.
This is called “conversational implicature.”
The idea behind conversational implicature is that what we say in a conversation is hardly ever literal so we subconsciously identify the real intention of the speaker by drawing upon the library of shared “world knowledge” that underlines all our conversations.
If both sides respect the cooperative principle and “read” conversational implicature, it allows for significant shortcuts during the course of conversation.
For instance, take the example given by James Giangola at Google. The two-sentence conversation below contains a wide variety of implications.
Grice further explained that for a conversation to feel natural and comfortable the cooperative principle also needs to respect what he called “maxims:”
Means that the speaker gives the listener only as much information as is necessary to further the perceived purpose of the conversation. But not too much. Did you ever feel someone talks too much or in unnecessary detail? This person was ignoring this maxim. So, always tell only what you think the listener needs to know.
Common Chatbot Error: Allowing your chatbot to share too much information in response to one conversational quarry. Giving the user a long chunk of text to read which feels unnatural in a conversation.
Is about what we say, the truthfulness and quality of information. A speaker who lies or misinforms ignores this maxim.
Common Chatbot Error: The information the bot provides is incomplete, incorrect and/or outdated due to poor planning or lack of maintenance.
Means exactly what you think it means. It’s about contributing only the information that is relevant to the topic of conversation.
Common Chatbot Error: In an attempt to push the sale, the chatbot offers information that is not relevant to the user’s question without giving the user a chance to opt in or out of receiving this extra information.
Expresses the way people attempt to communicate clearly, without ambiguity.
Common Chatbot Error: The information provided by the chatbot is unclear, ambiguous or contradictory leaving the user more confused.
Clifford Nass, a renowned authority on human-computer interaction claimed in his book, The Media Equation that “people tend to treat computers and other media as if they were either real people or real places.”
Hence, when presented with a conversation, people subconsciously follow the rules of a human-to-human conversation even if they are aware that they are speaking a software. They expect that software, your chatbot persona, to follow it too.
Another pillar of a functional conversation is turn taking.
Seems obvious, yet many first-time bot designers forget to give users space to actually interact.
If you have used a chatbot in the past, you might have experienced being sent a message after message without being given the chance to respond. If you are to have a conversation with the user, you must allow for it to happen. It goes hand in hand with respecting the maxim of quantity.
Peter Hodgson identifies turn taking as the mechanism by which we resolve ambiguity and repair conversations.
Chatbots are not sophisticated enough to understand subtle social cues, so the role of the designer is to make transitional prompts (such as questions) more explicit yet natural.
Design conversations that work for the users in their context.
Talking to machines is still in its diapers. So, in order to make people feel comfortable, you need to focus on the user’s physical and emotional context and the kinds of conversations that can serve them in that context.
- if you are designing a voice interface, you need to take into account the social environment of the user;
- if a user is on the go using a mobile device, you should design your questions and need for answers accordingly (keep it short & to the point);
- is the chatbot assisting users in emotional distress, in a hurry?
There are multiple conversational elements you can use to put the theoretical principles of conversational design into practice:
Bot introduces itself and its function.
Example: Hi, there! – Welcome! – My name is XXX and I will/can…
A natural end to a conversation to provide closure to the user and highlight the bots social intelligence.
Example: You’re all set. Thank you for your order. Goodbye.
One of the most effective prompts to keep the user engaged with the conversation, gather information and narrow focus of the conversation.
Example: What kind of topping would you like? – What’s your name? – What size do you prefer?
Ensures user that their input has been received.
Example: Okay. – Got it. – Thank you.
An informational statement can manifest as general information (statements answering questions), an overview (how the information will be structured within the conversation) or a menu (a list of options). Every information statement should be followed by another prompt.
Example: Here are the opening hours, would you like to know anything else?
Suggestions can be provided by your chatbot to help the user answer a question or make a decision that is within the power of your bit. You can also use them as hints to lead users to discover new features.
Example: If you want a free express delivery for your future orders, I can sign you up for our membership. Are you interested?
An important component that you should try to avoid using too often as it highlights bot’s shortcomings and can annoy the user. It should always be followed by offering an alternative option, it should not be the last thing your bot says.
Example: Unfortunately, I don’t deliver to your ZIP code, would you like to pick up your order at one of our centers closest to your home?
A form of user request. You should not have to teach the users what to do, the action should be clear through the conversational principles.
Right: What do you want to do next?
I want to keep shopping. – I want to check out.
Wrong: To continue shopping click/say “keep browsing” or select “check out”
Provide users confirmation on how the bot understood their input. It helps to eliminate anxiety and doubt in the process.
Example: You ordered 12” pizza with a cheese crust, pepperoni and extra onion. Would you like to make changes to your order or proceed to checkout?
Discourse marker linguistically/emotionally relates our upcoming words to what was previously said. It aids comprehension and makes the conversation seem more natural, fluid and less robotic.
Examples: Also – Above all – Ironically – Fortunately – As an example – In that case – In other words – Consequently – By the way – As a matter of fact – For that reason…
- By the way, did you know we have 2 for 1 promotion for all….
- In that case, let me schedule you an appointment with one of our sales team representatives!
Errors happen when your chatbot cannot proceed because the user didn’t respond, said something the chatbot can’t understand or requested something the bot can’t do. This may happen at any point in the conversation – the way you handle errors defines the success of the whole conversation.
Example: Sorry, there seems to be something wrong with your email address, please type it in again to make sure we got it right! – Sorry, how many tickets was that?
Buttons are a pre-set visual element of a conversational UI. They help users to focus conversation; discover possible next steps, related topics or conversational pivots; and take quick action. They can be written or visual (images) or a combination of text and image.
Buttons are the key element of rule-based conversational interfaces (see section “Select a type of Chatbot”)
Example: What would you like to do next?
Button 1: Subscribe me to the newsletter.
Button 2: I wanna learn more about the product.
Technology-enabled conversations allow you to use a wide variety of media as part of the conversation. Audio, video, Gifs and images can be used to answer questions as well as add personality to your bot.
5. Conversational Interface Design: Where to Start
Now that you are familiar with basic pillars and functional elements of conversational design, you can start thinking about your bot.
Though, there are a few steps that need to be done before you get to the technical part of creating device-independent and human-centered conversational flow:
The first task on your list is defining the audience you expect to interact with your bot.
Thinking of your users first will help you avoid designing the bot to fulfill your goals only.
It’s essential you do solid research at this stage as it will save you from making major changes when your bot is already set up.
You can start by asking and answering crucial questions:
- Who are your customers/users?
- What are their problems, needs and goals?
- How do they try to solve these problems and achieve their goals?
- Which words and phrases do they use to talk about these actions/tasks?
- What kinds of situations or circumstances trigger these actions?
- What is their context?
Answering these questions helps you form specific user personas – short descriptions of most likely (or ideal) individual customers.
Once you have the persona, you can define his or her customer journey – the pathway the customers follows to complete their goals. Naturally, a customer can arrive at your solution/brand/company using many different pathways. Your job is to identify those that are the most common and most important (to the customer).
Create 2-3 specific user personas and their journeys that describe your best customers.
Use real customer data, not just your impressions of customer problems and behavior.
When defining what use cases to target and what type of bot to use, you need to consider:
- Technical limitations
- Level of effort required
First, when it comes to conversational interfaces you can opt for:
AI bots leverage Natural Language Processing (NLP) and machine learning to communicate with users. They can exist as graphic interfaces or voice assistants.
These types of bots give their users more freedom of interaction and hence provide a level of sophistication rule-based chatbots can’t. However, they require high technical knowledge and more complex script writing. Ai-driven interfaces are difficult and costly to build and maintain and, due to more user freedom are more likely to result in errors.
Still, if you get it right, it can revolutionize the way you communicate with customers.
Recommended to: Larger businesses or businesses with a substantial budget for projects that are not time sensitive and have time to mature and develop properly.
Rule-based bots do not require AI to function properly but rather rely on the premise of “choose your own adventure” giving users conversationally designed options to help users solve their problems.
Non-AI bots give your users less freedom in their answers and so maintain you in control of the conversational flow. While less technically sophisticated than AI bots, the concept allows you to develop complex structures and flows with little or no technical knowledge. If well designed, they can be incredibly effective at a fraction of the AI bot cost.
Recommended to: Small businesses, entrepreneurs & marketing agencies with a limited budget and low technical knowledge that need to produce bots for multiple use cases promptly.
As per defining the role of your bot, the idea is to direct your effort where it will have the most significant impact. Start by listing scenarios (use cases) in which your customers would find the bot useful. Will you get the most out of the conversational interface when used as a first lead qualification tool, shopping assistant or customer support?
- use cases that affect the largest number of (potential) customers or
- use cases that make a considerable difference for a small number of extremely loyal customers.
Essentially, a chatbot persona – the identity and personality of your conversational interface – is what makes digital systems feel more human.
It’s there to give your customers a consistent experience that doesn’t feel like talking to someone with a split personality disorder.
This is not optional.
If you want to design a successful conversational interface, it must have a defined personality. Not just for a better CX but also because chatbot flows are often written by multiple people who will struggle without cohesive guidelines.
Power of personality can take you a long way.
Therefore, think of your persona as a character in the virtual world that exists in the same way as a character in a book or a movie. This character is an extension of your brand, hence it should have the manners, knowledge, and attitude of someone you would actually hire to face your customers. Also, it should be someone who is capable of connecting with your target audience so be sure to consider factors such as lifestyle and demographics.
Define your persona’s:
- Tone (cheerful or calm?)
- Level of politeness
- Writing patterns (e.g. does the persona use abbreviations?)
Google’s guide to creating a persona suggests this process:
- Step 1: Brainstorm characterizing adjectives. Focus on the type of qualities that can be perceived through messaging (friendly, funny, competent, compassionate)
- Step 2: Next, cut your list to 4-6 most important adjectives that will make up your bot’s core personality.
- Step 3: Define several characters that could embody this personality, it may or may not be a person (it can be an animal, robot or a fictional character, etc.)
- Step 4: Select the character that best embodies the chosen personality traits as well as matches your brand narrative and values. Focus on defining personality traits and things such as lifestyle choices or a language to which your audience can relate. Avoid specific features such as age as this usually isn’t a critically defining feature.
- Step 5: Create or find a visual representation of this persona. It will make it easier to remember and associate with. You can also use this representation as the bot’s avatar on the screen.
Regardless of how tempting it may be don’t start by writing the script. You can tune the linguistic and conversational nuances later, for now, stick with the practical functional version of what is to be said.
First, you need a bulletproof outline of the dialogue flow.
This outline will be the “skeleton” of your bot.
It’s an essential step from understanding human interface to effectively using the power of conversation and personality to lead people through the entirety of their customer/user journey without friction.
At this point, you should already have all the “ingredients” to create this flow. You:
- know your audience;
- selected the role and function of your bot;
- identified crucial use cases and scenarios.
Outlining the flow means writing down the questions in a logical sequence with all possible answers and follow-ups to those answers. This way you are likely to identify missing paths and dead ends and add them flow to ensure that the conversation sounds natural no matter what path the user takes.
In essence, you will be creating a branching diagram.
A linear conversational flow is a question-answer model which doesn’t give any options to move away from the main subject of the conversation.
A non-linear conversation flow allows for conversation to take various routes during the conversation including moving backward or stirring towards another topic. This, if designed properly can make the conversation sound significantly more natural but it is also much harder to plan. You will need to keep in mind that there will be multiple ways to reach one question.
In general, there will be 1-3 main paths within your flow a customer can choose to complete his or her goal.
However, remember this is only a structural overview. What will make your bot really work is a conversational designed derived from the way people talk and chat not write.
Once the flow diagram is in place, you are free to expand and perfect your script.
So, get to writing keeping in mind the wonderful bot persona you created earlier. You can start with the main flow and branch out as needed.
But, before you do, have a look below on the Do’s and Don’ts of chatbot script!
As Ruben Babu points out in his recent article, chatbot conversations need to be written in a way that helps users:
- Navigate through complex systems (Navigability)
- Find out what’s possible and what is not (Discoverability)
- Achieve their goals (Usability)
So, you need to keep in mind that you are designing a conversation with a purpose – it can’t be just pleasant, it must be easy and efficient to navigate.
Here are several fundamental Dos and Don’ts of conversational design:
You would think this is something fairly obvious, but it’s surprising how many first-time CUI designers let this slip their minds.
What does it mean being “conversational”?
Well, in essence, it’s about avoiding plain, impersonal statements you would never ever say when talking to another person.
Wrong: “The email you entered is invalid.”
Right: “Actually, there seems to be a problem with the email address. Try retyping to make sure I get it right.”
Wrong: Appointment scheduled for May 24, 2019, at 10 am.
Right: “I scheduled your appointment with one of our superb sales representatives for May 24, 2019, at 10 am. I will shoot you an email with the confirmation.”
Although it’s clear the user is speaking to a machine, it needs to feel like a living, breathing character.
Note: If you are designing a rule-based chatbot, you also have to think about providing users with conversationally appropriate answers that feel natural.
The fact that users are operating with buttons has many rookie designers tempted with using “button” language such as “Confirm” “Send” “Accept” “Go Back”
So, again: If you wouldn’t say it, don’t put it into your conversational thread.
Instead, design buttons that fit your audience:
“Why not, sounds interesting 🤔”
“Looks right to me 👍”
“No, I’m not in the mood”
“Please show me the options again.”
These two are basic conversational elements for a good reason.
No conversation ever starts out of the blue. There is always some form of greeting or initial pleasantry to get things started. Similarly, no polite conversation just stops without some kind of conclusion.
Would you walk into a flower shop, ask for help picking a bouquet, pick one, pay and then leave the store quietly without saying thank you and goodbye?
The shopping assistant would also try to conclude your interaction in a pleasant, conclusive way.
When we talk about “greeting and ending,” it doesn’t just mean saying “Hello” and “Goodbye.”
You can make your greeting more complex by pairing it with an introduction of your bot, its persona and/or its capabilities:
“Hello, my name is XXX and I am your personal shopping assistant.”
However, sometimes that might be too much. Perhaps a simple “Hey there, how can I help?” would be sufficient. As we mentioned when discussing conversational pillars ALWAYS take the context into account.
Speaking of pillars… don’t forget about conversational implicature. You don’t actually need to say “goodbye” to say “goodbye.”
“I was happy to help. If you need anything else, I’m always here!”
This statement implies:
- This conversation has come to a conclusion
- A goal has been achieved
- Goodbye for now
- I am here for further assistance if necessary
Designing in long chunks of text is another of the most common mistakes committed by first-time bot designers…
Remember when we talked about turn-taking?
You are writing a conversation, not a blog. If the customer wanted to read long explanations and description, they would visit your website and not talk to the bot.
So, when writing things down think:
- How will the text look in a messaging app or pop-up window
- Will the text fit on the screen?
If you must share more information, do the same thing a person would do in a chat, break it down into multiple bubbles.
Note: Don’t write more than 3-4 bubbles per turn as you risk losing the users attention.
Messaging universe is full of fun possibilities; possibilities that invite emoji, Gifs, images and videos into the conversation.
Emojis and rich media allow you to make up for the missing gestures and expressions we perceive in a real face-to-face conversation. Hence, creating an engaging interface or visual design has never been easier.
Use emojis to add a little lightness into the conversation.
Leverage rich media to substitute the tempting long chunks of text. Images and videos speak for themselves.
- don’t overdo it (it’s still supposed to be a conversation after all)
- always remember the context (are your users usually in an environment that allows them to watch a video, etc).
If you are designing a chatbot, don’t design it just for one channel. Strive to create independent, human-centered systems that will work on multiple channels.
This way, you will be able to implement and leverage a single chatbot on various channels and in various formats such as Facebook Messenger, WhatsApp, website embedding, chatbot landing page or even Google Assistant.
Your chatbot should always keep customers informed about what is going on by providing appropriate feedback.
Design your bot to
- Be clear and specific about the process – leave no space for ambiguity
- repeat important information (email, address, booking details, etc.)
- allow users to amend them if necessary
“I sent your request to our sales team, one of our members will get in touch with within 24 hours via email.”
“Hurray! Your order has been processed! You can expect the delivery within 4-5 working days. Oh and.,. Remember to check your email for confirmation and a shipment tracking number.”
“Just to be sure, please double check your details:
- Phone Number
- Date & Time of the appointment
Do you want to change anything?”
Conversational interfaces work because they feel natural and people intuitively know how to use them.
So, if you need to “teach” people how to use it, you are doing it wrong.
Wrong: “If you want to see more options click/type – more options – ?”
“Select one of the options below or click to see more options.”
Right: “Would you like to see more options?”
When constructing your thread ensure that every single branch has an appropriate ending and doesn’t leave the user hanging in a limbo.
There is nothing more frustrating than getting stuck and having to re-start the conversation.
Double and triple check that every thread is connected and/or has an appropriate ending.
Always, under all circumstances keep your chatbot persona in mind.
For instance, if you have established that your persona abbreviates (e.g. “you’re”) or has an accent that reflects in its writing than keep doing so throughout the conversation.
What goes for writing habits goes for the personality too.
If you established your persona is bubbly and quirky, keep it constant. If your persona is calm and compassionate don’t throw in a joke all of a sudden.
Seems silly and simple.
Nevertheless, it’s a very important step.
Do read your thread 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.