Conversational Design

Foreword

When I left

the east coast to work in the venture capital industry, I was struck by a number of realizations—east and west coast designers were doing such different things, and the tech industry had changed drastically while I’d been busy in academia.

But the biggest surprise of all was how disconnected the tech industry had become from the humanity that gave birth to it all. I’ve been lucky enough to have had access to a computer for forty years, and I’ve always fought to give the arts a first-class role in its evolution. Yet I found few humanist technologists working in Silicon Valley.

Erika Hall was the exception. Erika represented a point of view that was grounded in not just what the technology could do, but—more importantly—in what real humans needed to do. She was challenging the comfortable models of tech’s best practices, like ubiquitously produced personas—artifacts that gave the appearance of serving users, without necessarily teaching designers how to listen to them.

I like to tell designers that their job isn’t to be storytellers, but to be story-listeners—people who takes the time and energy to listen to others’ stories. That’s the essence of a conversation: people listening to each other. And a conversational design approach can help designers serve people better.

Conversational design is truly human-centered design, Erika writes, and let me tell you why that resonates with me. She focuses on what designers really need to know, including the idea that there is no simple solution when it comes to designing for humans.

At the core of Erika’s quixotic work is an effort to ground the designer’s aspirational goal of empathy in reality. Real people. Real feelings. Real conversations. And above all: real listening.

When designed well, conversational interfaces are able to amplify the human-computer bond—and create a relationship grounded in communication. When they work poorly, they produce mistrust. When they work effectively, they promote trust.

Trust lies at the heart of great design. And trust begins with listening.

—John Maeda

Introduction

We have all heard it said that one picture is worth a thousand words. Yet, if this statement is true, why does it have to be a saying?

—Walter Ong, Orality and Literacy

I once worked with a client

in an entrepreneurial corner of an old-school, multinational publishing company. Tilting towards the future, the publishing company had developed a web-based, semantic search engine. They hoped the search engine would help business leaders extract more value from news by making meaningful connections among stories.

From the moment the site launched, virtually every visitor to the homepage bounced without taking action. The creators were shocked and dismayed. They had believed the technology was so exciting and powerful that their audience of executives would find the intricacies of the interface delightful. But the design was a flop. So, they hired our design firm to fix it.

Untangling the functionality and simplifying the interface was going to take a bit of time. I suggested we at least add an instructive sentence or two to the interface in the meantime.

The Client: Well, our team has talked about that and we haven’t been able to agree on exactly what it should say.

Me: Why don’t you just put a few helpful words at the top of the page and you can change it as soon as you think of something better.

Oh, no. We couldn’t do that. We have to go through an approval process and we can’t just rewrite copy once it’s been approved.

But right now there are no instructions. No one who lands on this page has any clue what to do. Really, just a single sentence would be better. And you can change it at any time. This is the web. This is your interface, not a contract.

I’ll take your idea back. But we really want to be careful and make sure it’s right before we commit to anything.

My arguments went nowhere. Authority prevailed. We would just have to wait for the new design.

This story has a bitter end. The parent company lost faith and killed its own darling. They shut down the business unit and sold the technology for parts before we could get our work out into the world. Anyone who’s worked with a legacy business trying to transform itself by embracing digital technology has seen this sort of thing happen.

That seemingly minor objection to adding some language to the interface stuck with me. It was at the root of why the project failed. When a publishing company with established processes for writing got into interaction design, old and new approaches across disciplines collided and produced an extreme illustration of a common and often unexamined divide.

The Great Divide

The client confounded us by having two decision paths. Anything involving visual elements or non-verbal interface behaviors was on the design-decision path. On this path, they were comfortable trying things, working collaboratively, and iterating quickly. And they trusted our recommendations.

Decisions about words were on the writing path. These decisions needed to go through the established editorial process. The client fretted over every revision and expected the choices we made to represent commitments—once edited and approved, language would not change. The fewer words we used in the interface (and therefore the more it represented design), the better the client felt about our work.

Over the years, I’ve seen this dynamic play out again and again. I’ve worked with clients who asked whether we could use an icon instead of text, and interface designers who’ve broken into a cold sweat when asked about the lorem ipsum.

Organizations committed to multidisciplinary design collaboration will put together teams of visual interface designers, developers, and interaction designers, yet there’ll be one writer working in solitary two buildings away.

We’re all working on one system. Why do we form teams around the artifacts people produce instead of the problem they should be working together to solve? It’s not as though the user is going to interact with the information architecture, then the visual style, then the code, and, at some point after the editorial approval process, pick up the interface copy to read at bedtime.

Interactive digital design is still ensnared by its graphic design roots. Portfolios consist of sets of rectangles. And discussions about Graphical User Interfaces (GUI) tend to focus only on the graphical aspect. As much as people still talk about thinking outside the box, design often remains comfortably inside it.

Just draw a rectangle shape, maybe 5 or 6 shapes on a sheet of paper, then draw the features you want on each view of your app.

—Beginner’s Guide to iOS Development: The Interface

It’s easy to conflate the value of the system with the surface appeal of its visual representation. First make nice rectangles, then add meaning to rectangles. But rectangles of features do not represent the value in our apps. And there is no way of knowing what each view should include until considering what the user needs. Designing the container first makes even less sense when the containers are starting to disappear.

Our Robot Friends

If we believe machine intelligence will make our applications smarter, they might as well just start talking to us. That’s why conversation is the new interface.

—Roy Bahat of Bloomberg Beta

A confluence of recent advances in technology is radically reshaping how humans interact with digital systems—and with each other through these systems. Cloud computing has allowed internet-connected devices to access immense amounts of data storage and processing power. The popularity of texting gave way to the rise of messaging apps that are cheaper than SMS, faster than email, and make it easy to share all manner of media. Increased access to processing power and data also gave a boost to machine learning. This has allowed computers to perform tasks by generalizing from examples, and to improve over time based on new data—and all this without explicit programming. Apple put Siri in the iPhone. And Amazon made a talking speaker.

So, now there’s a lot of enthusiasm for chatting with software through conversational interfaces that can interpret natural language on the fly—whether typed or spoken. Chatbots text us and intelligent assistants answer our utterances.

It’s terrific that we’re now talking about the importance of conversation as an interface, and language as part of design. However, an interface that seems conversational on the surface may not be conversational in practice.

Conversation is not a new interface. It’s the oldest interface. Conversation is how humans interact with one another, and have for millennia. We should be able to use the same principles to make our digital systems easy and intuitive to use by finally getting the machines to play by our rules.

Unfortunately, overly literal interpretations of the idea are leading to systems that are hard to use. Being able to exchange text messages with a bot doesn’t necessarily make it easier for people to reach their goals. We must go deeper. Otherwise we’re just making things harder on ourselves and those we’re designing for.

Towards Conversational Design

These days, many designers and businesses are embracing the idea of conversational interactions to mean literally talking to or messaging a digital system. Advances in artificial intelligence, voice user interfaces (VUI), and chatbots show a lot of promise. But conversational design is more than creating interfaces that talk and text.

Taking a conversational approach to interaction design requires applying the deeper principles of how humans interact with one another so we can create systems that succeed on human terms, no matter the mode of interaction.

Fortunately, we