Designing for Dialogue: An Introduction to Conversation Design and Its Impact on Customer Experience

Nikki McLay, our Conversational Architect here at VERSA Connects, loves designing intuitive, human-centric AI experiences.  I sat down with Nikki to pick her brains on all things Conversational Design – and why this type of tech is emerging so quickly in the Customer Experience space.

‍

‍First and foremost – what is Conversational Design?
Conversational Design enables users to interact with and navigate a computer experience using natural sounding conversation as an interface. It’s both the design of the interactions that people have with the computer they’re engaging with, and the training of the computer, to learn to understand human conversations and become better and more natural over time. It’s a great combination of technology, UX design, copywriting and psychology.

‍Cool, so it acts as an umbrella term for those things you just listed?
Yeah!  

‍What is Conversational AI? Is it just the copywriting aspect of Conversational Design? Where do they overlap?
They work together. We design an interactive interface which feels like a conversational experience, and Conversational AI processes it. It’s a type of artificial intelligence (AI), that uses a combination of Natural Language Processing (NLP) and machine learning (ML) to replicate the way humans naturally communicate. These systems are trained on data which is processed in several ways, teaching the system how to understand and process human language. It’s constantly improving and learning from the interactions it has, becoming more ‘intelligent’ over time.  

Is it easier to use Conversational AI for voice or chat?
Both are unique. Text input has the benefit of also having visual aids, you can read and correct a typo for example, whereas with speech in more intuitive, there is more room for error due to how nuanced human speech input can be, but today, with good design, it can be close to flawless and will only get better. When you are navigating an experience by voice or on a chatbot, different machine learning techniques process your input. Speech to text is a complex machine learning process that converts our natural voice input into text which is then analysed; sentiment analysis is another process which looks at that text and determines what your intent is and decides where to graduate to next in the conversational flow. It can also detect the emotional tone of your words, which in turn helps the bot respond accordingly. Technology is getting better all the time, so having a natural experience is easier.

However, this is all quite dependent on how well the AI is trained to accurately translate the speech input to begin with and how well we designers engineer the flow of the conversation. We all pronounce and say things differently, so when designing our conversations, we are also thinking about how the inputs are going to be used effectively for training our Conversational AI agents. As architects of these systems, we have a duty to make sure it is exposed to a myriad of different voices, accents, languages, and expressions which will provide more rich and diverse data for it to learn from. If we fail to consider accents other than our own, or those with impaired speech due to disabilities, the harder it can be for the bot to understand them. This results in some people not wanting or being able to use something intended to make their live’s easier, so, it’s critical we make sure our agents have been trained with enough diversity to make the experience enjoyable and accessible for as many people as possible. This is an interesting challenge and poses the question: how do we design effectively when our primary audience could be anyone?  

‍That is super interesting, because I know that this type of tech is accessible in that it allows people who may be visually impaired to interact with things in a whole new, different way, but I didn’t think about how it can be limiting if you don’t feed it the right information to begin with! Therefore, you must expose it to heaps of different people living heaps of different lives, right?
Yeah! We’re designing for everyone—your biggest audience—whilst battling your own internal biases and assumptions about who ‘everyone’ may be. The nice thing is constantly thinking about how diverse we all are and doing our best to improve our designs by incorporating those differences into crafting better conversations that both AI and people will benefit from. I love just listening to everyone interact with each other as I go about my life and bring little insights back to the design process.

‍Even though this kind of technology has been around for a while, there’s is so much conversation around Generative AI and Conversational AI at the moment. Why do you think that is?
You’re right, it has been around for a while, but for a long time it just wasn’t as accessible to everyone yet due to high costs, a lack of understanding on how to work with it, among other limitations. For those of us designing experiences, it was also frustrating that the advancements weren't keeping up with the ideas that we were having. We were trying to do amazing things with it, but the barriers were too high to be able to implement them a lot of the time.  

Now that larger companies are releasing more accessible Large Language Models combined with Generative AI, people are of course extremely interested in how they can use it, as they can see real life examples now. Also, developers and designers can experiment much more easily, bringing more people and therefore fresh ideas and innovations into the space.

This kind of technology is having a pretty huge impact on the Customer Service space, which is what we work in – why do you think that is? Why do you think it’s something people have decided fits in that space?
The foundation of Customer Service, particularly in the call centre industry, is the concept of wanting to talk to someone. So Conversational AI is perfectly aligned for this space.

The chatbot for example has been around for a while, and everyone is well-accustomed to those little boxes popping up and understands intuitively how to interact with them. It’s easy for us to understand that they are designed around this core concept that you can have a chat about something and get an answer.  
But often the experience feels quite limited and unpleasant.

With Generative AI, Large Language Models, and Text-to-Speech AI becoming more advanced and more accessible, I feel that this has allowed people to realise that they are not so limited anymore! When it comes to Customer Service, the prospect of being able to offer elevated Customer Experiences with more natural sounding, intuitive assistants that use a combination of these systems in a creative way, it becomes not only a pleasing idea for both businesses and customers, but one that makes perfect sense.

‍Yeah, it’s exciting that those kind of experiences with chatbots that were traditionally not that amazing, are getting the light of day to become way better, and service customers way more efficiently – which is great! I feel like everyone has probably had an unhelpful experience with a traditional chatbot at least once.
Yes! What we are seeing now has raised the bar in terms of what a good Customer Experience can be. Even humble old chatbots are far more impressive and interactive already. AI took a big leap and it was like going from a Nokia to an iPhone, and it’s opening a lot of real opportunity in the Customer Service space.  

‍Do you think that the acceleration of that technology was always humming along in the background? Because lately, it feels like it’s been exploding.
Definitely. Unless you were following it closely you maybe wouldn’t know it! But then when it begins making the headlines every day, you’re like wow! This must be new. But a lot of very clever people have been busy in the background for an awfully long time working on not only Large Language Models (Eg: GPT3, Claude, Bard etc) but all the other kinds of ML including Generative AI. The combination of everything working so well together is something new and exciting and, in my opinion, right on time.

Do you have any words of advice for a business owner or contact centre manager who is considering introducing Conversational AI and Generative AI into their operations?
I would encourage them to embrace Conversational AI with a mindset of innovation by design. When you integrate automated conversations into a team that is already conversational by nature, there’s the opportunity to synchronise technology with your own work stream to really make a difference with more productive, less stressed teams able to achieve high standard of customer service easier in a very cost-effective manner. Embracing new concepts and gaining knowledge early on fosters familiarity and understanding, which will become a big advantage in the long run too. So, rather than falling behind, now is the chance for businesses to lead the way: to continue to innovate and stay ahead of the competition.

‍