Podcast | How to get your chatbot to emulate your brand personality? Here’s how.


Director, Strategy & Content | Podcast Host of Mobile Interactions Now

22 min podcast
tyntec Podcast with Marco Spies

In this episode, we check in with Marco Spies, Co-author of Branded Interactions and Co-founder of think moto, a Berlin-based design and innovation company. We’ll be talking about what Marco learned from helping some of the coolest brands create conversational experiences for sales and support. (In-car, too :-))

Podcast transcript


Marco, welcome to the show. I am so thrilled to finally have you on the show, because this is the 25th episode of the Mobile Interactions Now. So far, we've been dealing with a lot of technology, and I really wanted to talk about this pressing topic of brand strategy and design.

Part of the reason being, now that technology is catching up and providing a lot of components to work with, the brand side of the conversation seems to be lacking now. And it's becoming important, as we are looking at digital interface as really, a primary experience touch-point for many brands now. Before I get totally lost in that, tell us a little bit about yourself and what you do, and then we take it from there.



Yes. Hi, Jean. Thank you, first of all, for having me. Yes, I am Marco Spies. I'm the CEO of think moto, a design and innovation company based in Berlin, that I co-founded 10 years ago, together with my partner Katja Wenger. And what we do is innovating brand and product experience for companies, quite large companies like Volkswagen, Audi, Lufthansa, and the likes. But also for quite a broad range of German SMEs, the so-called German Mittelstand. And a number of startups from the Berlin ecosystem, but also from Russia, from the US. Currently from the UAE.

Maybe it's important to add that one of our focuses, as a design and innovation company, for the past three years has been in creating conversational experiences. We have created chatbots for sales and customer service automation, but we have also been working on the Audi in-car voice assistant for the future. So this is like a concept study for the future.

Recently, I may add this as well, Katja and me, we have been co-founding another company, called IGP, where we develop a platform for conversational commerce as a service.



Wonderful. Especially, in this new reality, pandemic reality that we are going through. I do believe that digital interface is becoming the first touchpoint, and throughout the entire customer journey, this permanent touch-point for many of the consumers, especially.

Tell us a little bit about the state of the art, where we are at in terms of making it both meaningful from the brand's perspective, strategically, as well as delivering fulfilling, satisfying experience for the users.



Yeah. First of all, I totally agree with your analysis. And especially, digital conversations can play a lead role in a new game that is now happening. But I think we have to go a step back, and start with a change in perspective of what brands are and what they can be. When we ask how brands make their digital experience work, we already assume that a digital experience is something that can be viewed separately from other communication means, but it is not.

I think it is important to understand that the digital experience of a brand, is as much part of the brand as any other aspect. It actually is the brand. So the way a user experiences the brand will always be related to the brand perception, overall. And the same goes for conversational experiences.

Whenever a company starts developing a website, they know exactly what to do today. They know about user experience. They know about how to place the brand. But when today, a company starts a conversational project, they seem to have forgotten about all of this. I always say, they start like it's 1997. If you look at the website of Apple in 1997, you will see a website that has nothing to do with the brand as we know. But that was '97. There was really a technical journey to set up a website.

And if you look at Apple's website, 10 years later in 2007, you will see that there's everything that we associate with the Apple brand. There is the white space. There's the photography. There is the copywriting. There is the shadowing. There is everything that makes Apple, Apple as a brand. And you will also realize that between 2007 and today, which is 2020, 13 years later, nothing has happened in the website experience. So it's actually, websites have reached a plateau where they can't go beyond, in terms of brand experience.

Now, we have a new channel, which is conversational experiences. And this new channel allows us for much more intimacy and directness. When a company starts with this, they start as if there hadn't been all the experience in branding and user experience, and so on, with other mediums. And that is something that makes us actually say, "We have to change the situation, as an innovation company and brand company."



I hate dating myself, but I know what you mean. Back then, the first generation of a corporate website, that had no almost-tactile feeling that you experience now. Even though it's still a screen experience, it's all that. But you're mentioning this conversational experience that is being added to this. And I'm all for not having to repeat, and relearn what we learned when websites first came about, and go around and take that much time. Because what we don't have now, is that much time, and everything is happening incredibly faster.

A lot of conversational experience, yes, there are some live chat interactions happening on a corporate website, all those things. But for the most part, the website is technically an owned property. You own that channel. You can put your branding touch to it. But when you are working on, for example, using WhatsApp or other chat app, other social platform, and bringing that conversational interaction experience to that platform. Now, to me, that changes a lot of things. What you can shape, what you can own, how much of a strategic execution you can do. It doesn't sound like an easy task, so how do we even approach how to think about this?



Yeah. What you said is right. We do not have that much time. And we do not have the time to reinvent the wheel completely, because user expectations are much bigger than they used to be, because they know how digital products can actually work. So we better do it right, right from the start. And one thing that companies are lacking or missing, most of the time, when they start with this new conversation of channel, is to get to the brand right in that channel.

What you are saying is something that comes additionally, on top. You are usually not in your own space, but you're in a co-branding situation, somehow. But this is, to be honest, not really new. Because if we look back at the time when the apps came up, and whenever you are in a context of Google maps or social media, and so on, you're in a co-branding situation, as well. You usually have a multitude of brands.

If you have your phone in your hand, and let's say you're using Facebook, and you're looking up a brand on Facebook. You may have, let's say the brand of the phone, then you have a layer of the operating system. You have the layer of the app, which may be Facebook. You may have even something like maps within Facebook, and then you have your own brand. So you have a multitude of brands. And the same goes for things like Alexa, and so on. You are in a co-branding situation.

But what is important here, the thing is, what you really have to make sure is that your chatbot has a personality, and has a brand-related and brand-relevant personality. This might not be identical with the personality of the brand. But you definitely have to know about your brand's personality, in order to define a relationship. Really, like they were... How you say in English? Kinsmen or relatives, and they belong to the same family. And the personality of the bot has to have the same DNA, basically, like the overall brand.

We at think moto, have developed a branding model, already almost 10 years ago, which is called Brand BIOS. And this branding model, at the core of it, it's actually the brand meaning. Today, everybody speaks about purpose. That's what it actually is. So company, first of all, of course needs the purpose and needs to know what it is there for. But the second aspect that is really important is behavior. And BIOS actually stands for behavior, image, offering and story. These are the four aspects that we do have to understand of our brand, in order to create any brand-relevant digital application.



Is there a better way to get everybody, who are having different functional roles, to talk about strategy? To create, like, okay, so this is our brand DNA that we've been talking about. This is what we are, and this new chatbot is an interface that leads to it. How do enterprises, companies deal with this?



Well, the first thing is you start with imagining bot as a person with a strong, attractive personality that people can relate to. If you have defined this, it will be really differentiable on any third-party platform, to just get back to that point. So we conduct a one-day workshop where we define, together with the client, we define the personality of the bot.

From this personality, we derive behavioral attributes. That is important. These attributes have to describe the behavior of the bot. Then we cluster them. We create something that is called brand filters, and we define those brand filters for the different aspects of the bot. This is the language, but also, the overall tone of the brand articulation.

Also, visual aspects. Because talking about chatbots, most of the people think of these little windows in the right corner of your browser, that open up when you click on them. And then you have something that doesn't understand you. But this is not what we mean, when we talk about chatbots. Chatbots are actually, can be really multi-medial. Can include rich assents, and all this. So we have to think about that, as well.

And once we have defined all these attributes, we start writing sample dialogues. So there's certain situations in bot design that you always have to deal with, like a welcoming situation, like a good-buying scenario. We will define these in that workshop situation. But this is mostly to get the client's input and also to get the client's buy-in. We have to say at the same time, after that, we are working with a screenwriter, or someone who is really, really good in writing dialogue. What he does is he creates situations, based on the use cases that we discovered before, together with the client. The personality aspect is only one.

The other aspects are really to deep-dive into your user's needs, and do all the UX stuff that you always do when you build a website. We gather all the user tasks and so on, and then we define the use cases or user stories. Based on those, the screenwriter will actually create situations of conflict. That was something that we learned, that I wasn't used to as a designer. I studied language, actually. That is where my love for chatbots also comes from. But I wasn't aware of the importance of this, until we invited the screenwriter to work with us, to think of conflict in situations. Especially, in this sample dialogue.

What the writer would do is create dialogues, based on conflict. And in that, by using conflict-rich situations, you will learn much more about the behavior of your bot. When we created the Audi voice assistant, we created a situation where the voice assistant actually contradicted to the driver. And the driver, in another situation, started to make fun of the voice assistant.

So there were conflict situation, and we could learn to what level can the bot really live this personality? It's personality to also have, really, character. The bot with the most personality up-to-date, is still Kit from Knight Rider. That's a bot. That bot has character and has personality, but it may have too much for a car company, to implement something like that in a car.



I still need to play with this, because I was complaining how the BMW voice interface is very formal. It's like, wie kann ich behilflich sein? And just lacking some playfulness, that would make me want to talk back and things like that. But I know we talked about first-generation website and all that. But when we think about one of the earliest, at least in American context, examples for people to realize how far we came along is, do you remember Microsoft Clippy?



Yeah, sure.



That paper clip? And I was actually in Las Vegas when Microsoft was announcing it, way back when. And it was just completely a flop. But here's an example, it's such a character, because it didn't exist before. It has personality. Unfortunately, it became such a goofy, not so positive, not satisfying experience.

I mean, this day and age, I will be infuriated if I have to actually need that kind of assistant, virtual assistant, to figure out how to use an application. But back then, that was the reality. People were getting lost, what those menus are doing, and this assistant was designed to do it, but not to a success.

So to me, when you talk about those experiences, some kind of a intangible interaction, the exchange that is happening between the user and the bot, some kind of a system. Fundamentally, what has changed, that shows promises this experience will be a lot more satisfying for both ends?



One thing we learned is that we need something like a mood switch. Especially, for something like an in-car voice assistant, whose task it is to guide me and to also, guard me somehow. And prepare myself of critical situations. It needs to speak in a different voice than when I'm just having a chat with the bot, which should also be possible, because we want that users engage with the bot and speak to it. And so, that also the bot, and this is also from a bot design perspective, important, the more interactions we have, the more we can learn about the users. And the more personalized the interaction can be. Moods, which is maybe one way to deal with it.

You have something similar, actually, when you build text-based chatbots. You can guide, very well, through quick-route clients. We built a chatbot for lexoffice, a German accounting software. It was initially created for a campaign, Black Friday campaign, which should last for six weeks. In the end, it was so successful that it is still online, and it has now grown from sales to customer service. We build really strong UX funnels, and we had four or five UX tracks, different tracks. And we guided the users, once we had separated them and recognized them as a certain user-type, we could lead them with quick replies through the funnel. With this bot, during the campaign time, we had a misunderstood rate of 3%. That means only 3% of the user's intents were not recognized.



That's a fantastic result, I must say.



The key to success were actually two things. It was the UX, and the other thing was personality, again. It had an upper [inaudible 00:19:12] that would move. We had a full-spring web chat, actually. It was on Facebook messenger. It would react to certain intents by jumping around, or smiling, or crying, or whatever. It had really emotions that you could see. I would say, these two or three things are really important for successful bots.



When you design, you know how a lot of things come from the persona, understanding what we as brand experts and the designers think the persona of the user will be. Has this changed with the availability of data?



Well, for conversational, it's still hard to say, because there is very little data so far, about user behavior in conversational environments. From my perspective, the craft of UX is actually pretty much the same in conversational, than with any other digital medium. And yes, we segment our target groups. We built personas and we take them through the journeys, and we identify their user stories.

I think it's important to base personas on user research, on real research. By now, most of the time, most of the projects that we work on, we actually do speak with real users before. So we can inform personas really, by real user research. For bots, it is more important than for others to continuously track what's happening. How do users actually interact with your bot? Especially, since we have so little knowledge beforehand. So there is not that much standardization, like we have in web design.

We have quite a lot of freedom for design, and trying and experimenting, and doing things and learning instantly, and improving immediately.

Part 2 of the interview with Marco Spies will be released in two weeks, following this release of Part 1.