3 Secrets to Conversational Design
Reading Time: 13 minutes
Podcast: Digital Conversations with Billy Bateman
Overview: Our host Billy Bateman and Guest Hans van Dam discuss tactics to designing conversations that meet the rational and emotional needs of your user and training effective personnel by educating them in technology psychology and language.
Guest:Hans van Dam– is co-founder of Robocopy and the Conversational Academy. His goal is to develop and promote the role of the conversation designer. He develops curriculum for the Conversational Academy, lectures at multiple universities, and speaks often at conferences around the world.
Reference:
Host: Alright, everyone, welcome to Digital Conversations. This week my guest is Hans Van Dam. He is the co-founder and the CEO of Conversational Academy. And we’re lucky to have him today. He is going to teach us a lot about designing conversations, the future of the industry. And happy to happy to have you Hans.
Hans: Well, thanks for having me billy.
Host: So before we get into it, will you just tell us a little bit about yourself and how is Conversation Academy and how it came to be?
Hans: Yeah, sure. So it’s actually like when I was in university, I wanted to be a novelist. So I was a writer and I was just doing that. And I thought I was going to win a Nobel Prize for literature at one point. It turns out that that’s really difficult.
So I wrote a manuscript to the publishers, he answered, that’s not going to happen. So I ended up becoming a copywriter and I worked at a startup incubator here in Amsterdam that’s run by the university here.
But they had a program with Stanford and M.I.T.. So I worked there with a bunch of tech companies. And there I sort of got to learn more and more about technology, which was quite interesting.
And at one point, I hopped on board with what, a startup there which went bust. So then I had to get a job and I went to customer service real quick. So I got a job in customer service, actually doing customer service via social media. For one of the airliners here. It was quite fun.
And then I saw, you know, these chat bots pop up everywhere and the market was shifting from this traditional Q&A, virtual assistance to more conversational interfaces. And I saw that everybody was struggling with that. For me, it’s like, hey, I know how to write dialogue because I wanted to write fiction. I understand the technology. And now I understand a service space for me that was kind of like a no brainer to explore that.
So start freelancing. Then I met a bunch of other guys that have a behavior design studio. So they’re psychologists. And they were already using dialogues to make journeys easier. And they didn’t even call them chat bots. They had never heard of that. They were doing the same thing, kind of designing conversations.
So we teamed up and it made a lot of sense because we figured, you know, if you’re going to have a conversation between a bot and a person, then one has an artificial brain, and the other one has a human brain.
So we need to understand technology, psychology and language. And in that group, I sort of represented the AI brain and they were the human brain. So that was perfect. And we launched initially just being like an agency, you know, robo coffee. And we were in a conversation to that studio. And then I once made a mistake of writing a blog post that said that we train our own people in our own academy.
And that’s when everybody started reaching us from all over the world. That’s like oh God. Now we’ve got to start an academy. How do we do that? So we developed a program for chat bots really. So that was good fun. There’s a lot of interesting stuff already in there. Then one of the larger search engines, I guess, got a hold of us and they were like, OK, we need to train voice designers as well.
Can we help you sort of make the curriculum ready for voice? So that helped us with that and connected us with other designers. And that’s how we put together a program. And it’s really shifted the business now. And while we are training institute first and foremost.
And what we do now is, is really recognize, develop and promote the rules of the conversation designer. Like it’s a new job. It’s an important job and there’s really certain capabilities that you need to develop a certain skill set you need. So we tried to be that institute that sort of creates a benchmark for that, where we want arts certificate pretty much to be the scrum of conversation design. Everybody goes near a chat bot needs to get a basic training and understand the design process to create better experiences.
Host: Yeah. So when you’re taught your program, if somebody goes through it, they can essentially say, look, I know how to design, design chat bots and conversations.
Hans: Yeah, exactly. It’s really a step by step process that takes you from, you know, designing a bot personality for consistency and trustworthiness. I’m likely liking this off a chat bot interaction. And then there’s a design process where we look at, you know, the block needs and the user needs.
So the way we see the world, if we’re gonna have a conversation between a human and a bot, then they both need to get something out of that conversation. So we have a canvas that you sort of just fill in and you look at the user needs because the user as rational needs, but also emotional needs, like certain motivations, anxieties, certain expectations. So you learn to map that.
And then from the bot’s perspective, you know, the bot has certain information that it needs to give, certain questions that it needs to ask, certain responsibilities. Maybe it has super powers where it can look stuff up in a database real quickly. So you map that out on a canvas.
And then most important part in this process is sample dialogue. That’s the next step. Where you gonna get to people? And it’s super awkward. But one is going to be playing the user and the other one’s going to be playing bot and you sit them back to back from each other so they don’t have any visual cues when they communicate. You just let them role play. It’s like in prof- Theater.
And what this does is like, you know, the user. It’ll be super awkward to messy at the first try.The user will ask questions and stuff and they’ll go over it. And it helps you find the most natural flow of the conversation and understanding that process of how that works, of how a natural conversation is actually being created?
That is fundamental in understanding conversation, design. Because you see, I think you see this with a lot of your clients broadly. Right. You know, you start with technology or you have too much focus on knowledge management or, you know, it’s like they take the business process, turned it into a flowchart, and then you might ask someone to put some words out there.
Yeah, but that’s not how people communicate. So this exercise allows us to. So if, you know, get match go communication in there, and what happens is that when the user asks the question during a roll play, we sort of, you know, you have to think, is this a question that more people would have? Because every time you ask questions like you, you misinformation, you have a certain concern, a certain anxiety.
So we need to address that. And then the next iteration, that’s probably something that we proactively communicate. Right. So during a roll play, if somebody freezes up and is like that, how much is that? Right. And that kills the entire conversation. You know that that’s something that you need to address during the design.
Host: For sure.
Hans: People need to understand that better and get familiar with that process and, you know, feel comfortable it.
Host: Yeah, you totally spot on with that. I know you presented part of this at a conference a few months ago, and one of my team members was there and he came back. Now, whenever our team rolls out a bot or any client, part of that, QC, is they have to sit back to back and go through this conversational flow with no visual cues, make sure that it actually flows and it works before they role anything out.
And I know it’s a big difference. And that’s one of the great things he learned coming, that was probably the number one thing when he came back, it was like, okay, what you learned? Is like this guy, he showed us this exercise and it was amazing. So with being a conversational designer, like what is the skill set that you think somebody needs to do that and to be successful?
Hans: Yeah. So, I mean, you obviously need to understand some of the technology, some of the psychology and some of the language right. And therefore, it’s very rare to find all those different skills in one person.
So the more creative ones aren’t the technical and the other way around. So in smaller companies, show you just find the right person that can probably pull it off a little bit with some training and coaching from our end.
But at larger companies, we already see, you know, a bunch of different roles. We sort of see a conversationally eye trainer, that’s a bit more technical that deals with intense coping and optimizing and implementation and stuff like that. So those are people that are a bit more rational, bit more analytical than you have the conversation designers that will focus more on that role play and sample dialogue and validating that with users.
So these guys come from your UX department, these are UX guys and they can write a little bit, but they’re still not great writers. And at the top, you’ll sort of have your conversationally AI expert or strategist that is more focused on brand personality, on the psychology, on really all the different design components and conversation design. So he’s like your editor in chief. So we already see like three different roles early emerging in this field.
Host: So if I’m a small company and I’ve got one guy or a girl on my team that, you know, I think okay Suzy, she can probably understand the technology enough and is good enough writing copy. I want to plug her in. She’s doing OK. if they want to take it to the next level that would be like going through a program like yours could really enable that person to do it well.
Hans:Yeah, definitely. There’s just you know, there’s a lot of people everybody can type and therefore everybody thinks they can write. Right. And the thing is with because everybody can type, everybody has an opinion on text and it’s really designed the way we structure this is. The key is just design components that you need to use when you design these dialogues, you know, acknowledgments, implicit, explicit confirmations.
How do you write a prompt? How do you use turn taking? There’s just very fundamental techniques that need to be included in that design. So it is, it’s not just winging it and trying to create something nice. There is good design and there’s bad design. So it’s it’s following this course at our, you know, any one’s course doesn’t really matter.
There’s a right way of doing it and a wrong way of doing it. I think that’s what we need to educate the market on. It’s a serious job and it requires serious training. And just going through the course isn’t going to make you an experience conversation designer. Right. Yeah, you can get your driver’s license, but it probably takes a year for you to be a very good driver. It’s similar to that.
Host: OK. OK. Now, what awesome of the techniques that you would advise somebody to use if you’re building the bot to get people to actually engage with that bot?
Hans:Yeah. So, I mean, if you, what happens when you do the sample dialogue? Right. What that enables you to do is have much more empathy in a beginning because you start to understand your users so well. Right. Maybe you’re on a landing page and you want people to download your paper.
Right. So, you know, you go through to canvas and you start to understand your users. You look at, you know, the boosters, the barriers and the information meets that they might have.And the person getting into character is going to play that out. Right.
And the bots will, you know, get to understand that better and better and better. And here’s the thing. If you do as a user or as a human in general, if you feel understood, they’re okay with the other person taking control of the conversation. So if you feel like, hey, I really understand what’s happening in your mind, what your concern is and you sort of feel like I take you seriously.
Then you’re OK. And you want to follow along. Usually these chat bots are bad because in that respect, there’s no empathy because what happens when you don’t feel understood? That’s when you start talking a lot. Right. And that’s when you’re giving me input. And it breaks. So what we do is we have more empathy in the beginning.
The user feels understood. And then we can pretty much have a very simple dialogue with very binary questions, very short and simple.They’re usually a bit longer than other people’s designs, but they work better because you’re OK and you have trust in the experience. So empathy in the beginning.
Then there’s like, you know, short, simple questions that are very binary that you’re happy to answer. And per prompt what we really what we look for or per note is, you know, don’t make it too long. So we have to one breath test. If it’s you know, if you can’t read out a prompt in one breath, then it’s too long.
Then we have what we called a Jenga technique, you know, like Jenga? That way you have to tower with all the little blocks, right. You have to take them out one by one. So we do that with everything we write.
So we look at the tower of meaning and that we take out the words without the tower of meaning collapsing and just go over your writing. You can pretty much take out 50 percent of the words of everything that you write.
So you can make it much shorter and much snappier. So once you have the structure of the conversation and you’ve really trimmed it down, now it’s time to apply certain conversation, design fundamentals. So every turn that you write for, you want to have an acknowledgement, you want to have a confirmation in there. Sometimes implicit, sometimes explicit. And you want to end with a short question, pretty much a prompt to sort of let the other person know that’s their time to speak again. Right.
So once you have those fundamentals in place, that’s when you go sort of to your personality. Right? You want to make sure that the brand is represented, the boss personality that you’ve crafted is in there. So, you know, you have standard vocabulary that you do and don’t use.
And then the last step is really to apply more psychology. Right. So that’s where we get fancy. So if I have to ask you five questions, I can just ask you five questions.But I know that question two or three, there’s a good chance that you’d want to drop out. So what can I do about that?
Well, I can use expectation management and social proof. I can say, hey, I’m gonna ask you five questions. For most people, this only takes a minute. And I’m using expectation management and social proof. And that increases motivation. Right. So once you have the fundamental design set up, you can now start applying these psychological tricks to increase engagement, to increase completion rates and to just create better experiences.
Host:Wow. All great stuff. That’s something we’ve seen with the expectation management. You know, our team works a lot in drift, in intercom and you say we built a bot to book demos for a client. And if we tell them, hey, I can book you a demo, but I need you to answer three questions or four questions first. The completion rate is much higher than the bot just says, let me book you that demo. Then just ask, here goes question one.
Hans: Yeah. And if you if you would even present a reward more clearly in the beginning as well, then you know that also helps. And I can bring you a demo so that you can learn to save costs and live a happy life. [crosstalk].
Host: OK. That makes sense.
Hans: So what you generally try to do, if you understand the anxieties of the user and the motivation, what you do is proactively take away the anxieties and double down on whatever it is that motivates them. A Sort of keep that in front of them the entire time as you guided through the dialogue.
Host: Okay. This is this is all great stuff. Okay, we talked about the one breath test that was one another thing that trace brought back to us. So if I am a marketer. If I am somebody working in customer success and I’ve got a bot inside of my piece of software, I’m trying to drive adoption and usage of. If you had one piece of advice for somebody building bots. What’s your one piece of advice to them, to do it well?
Hans: To lead with design. I would say, you know, I think. If you’re an engineer, everything is an engineering problem and engineering problems are usually more expensive than design problems, particularly in conversation design.
If you figure it out, it’s just, you know, oftentimes the solution is just one or two sentences, right? So, yes, leads with design, lead with the conversation. So before you build anything, you know, do sample dialogue, you know, just use the canvas. Figure out what’s going on. Figure to understand the situation, play it out, validate it through Wizard of Oz test and talk to it. You know what Wizard of Oz test is?
Host: No, explain that to us.
Hans: So what you’ll do is like you’ll do a sample dialog that creates. It’s like you’re a little conversation. Right. And I know you end up with like a flowchart of post its on the wall. But still, that’s what that’s one person. It’s well-crafted. But you know, that’s just one person. So then you bring in about five to ten people that haven’t been involved with the projects.
And you just you know, let’s say we’re on a website, you want people to schedule a demo, then you know, you just tell them, hey, you’re on a website about this and that topic.And then this bot pops up. And then you just read out to prompts that you’ve written down on your post-its and you very quickly validate them. So you just have the conversation with them, right? So you say you let them say whatever they want to say and you just reply whatever it was that you have designed.
If you did it with five to ten people, you got because they’re going to say stuff that you have never thought of. They’re gonna be like, oh, god, that’s a good point. And then with five to ten people, you’ll discover everything that’s not working in your bot. And, you know, you can make an assessment. This way we will hand it over to an agent or fall with them to a Web site. Or is this something that we wanted designed for?
So we want to make sure that you got about 80 percent of your traffic go through 20 percent of the dialogue. Right.And then the other stuff you just hand over. So the Wizard of Oz test really helps you validate your design quickly. Understand what you’ve done to build and implement. So you gonna save a ton of money that way.
Host: Awesome. That’s great advice. Well, thanks for coming on. We really appreciate it. And if people want to get a hold of you and learn more, how could they do so?
Hans: Yeah, I guess the easiest way is, you know, conversationalacademy.com. Check out our stuff there. We just set up a nice little promo code for your audience if they use digital conversations as a promo there, on the conversationalacademy.com. They’ve got a nice little discount and can check out some of the courses. So that’s one way. Also I’m on LinkedIn and Twitter. Just reach out to me. I’m easy to find.
Host: Okay. Well, thank you Hans, great stuff about designing bots. I hope everybody implements it. I’m going to go back to our team and make sure we’re doing some of these things that Wizard of Oz test. It’s something we really haven’t done, but it’s something I think we’re going to start doing. So thanks again. And we’ll chat later.
Hans: Alright, thanks for having me. Be in touch.