Understand The Conversation Design Process

Learning Objectives

After completing this unit, you’ll be able to:

  • Explain the conversation design process.
  • Identify which stakeholders are relevant to conversation design.

All About the Process

Conversation design is being strategic and systematic about what you say and how you say it. It takes thought and planning, collaboration and effort. It’s not just writing dialog. It’s a process. In this unit we look at what that process looks like and who is involved. We also explore your role in the process.

Ready? Let’s get into it.

Here’s a timeline of the basic conversation design (CxD) process.

Illustration of timeline with four stages: discover, ideate, prototype, and revise.

The length of time spent in each segment will vary depending on your particular situation. Just like any other project you’re planning, the more complex the situation, the longer the process is likely to take. In fact, the CxD process on the whole is a lot like any product design process. It breaks down into the following basic segments.

Discover: Gather requirements from stakeholders, conduct relevant research.

Ideate: Create a document that lays out what your conversation aims to accomplish, and how you want it to look/feel/sound like for your users.

Prototype: Produce the information architecture (IA) in a flowchart showing how dialogs fit together, what menus will look like, and what voice and tone you’re going for. Craft the discourse and copy for what the bot will actually say to users in text, then assemble in a prototyping tool or environment.

Revise: Conduct usability testing, review with stakeholders for final approvals, conduct QA in final build.

Who You Work With

Conversation design is a team effort. Again, CxD is no different than any other design project in this respect. You wouldn’t set out to revamp your company’s entire product line or branding and messaging strategy all by yourself—unless, of course, you’re a company of one—so don’t try to tackle CxD alone, either! 

A good place to start is by figuring out who your relevant stakeholders are. You need the right people to help build, review, and implement the awesome new conversation strategies destined to delight your customers. For our purposes here, we sketch out the stakeholders in a typical enterprise-size environment. These are the folks who have a vested interest not only in how conversations shape your users’ experience, but also how they come to be. 

Conversation Design Stakeholders

Documentation, content strategy, content experience: These three positions are sometimes combined in the same department or even a single role. They serve as the interface between marketing and the actual copy creators. 

Subject matter expert from the domain of conversation (for example, service manager, sales leader, and so forth): Work with a subject matter expert (SME) from whatever vertical your bot is being produced for. Your SME’s experience working with colleagues, customers, and other stakeholders in their domain can help you hone in on an authentic voice and tone for your bot. For example, a service manager is familiar with how customer service reps talk to customers and one another. Use that domain knowledge so your bot uses the right terminology, lingo, and turns of phrase to feel authentic to your users. 

Marketing, User Experience/Design, Legal

Marketing: The marketing team can ensure that the bot’s style is in line with your corporate brand and voice.

UX/Design: They have the understanding of how to express your corporate brand graphically and where to situate the bot within the broader experience. 

Legal: These folks make sure you don’t break any laws or put your organization in legally fraught situations. Legal can make sure you don’t promise the users anything you can’t deliver (for example, “I can absolutely take care of this for you!” when the bot may not actually be able to, which can pose a legal risk). They also ensure you don’t ask users for information you’re not allowed to ask for (for example, age/health of employees, personally identifiable info of students). 

Engineering, Project/Product Owner, Systems/Tools Admins (Ops)

Engineering: Most bot builder software offers basic tools, but usually you need engineering to help you breathe a little extra life into your bot. For instance, If you’re working with voice, you may need engineering to customize the way the voice sounds (pitch, rate of speech, and so on). 

Project/product owner: These are the folks tasked with making this whole process happen by coordinating efforts across engineering, marketing, and other teams in your organization.

Systems/tools admins (ops): These folks often have knowledge or systems permissions to do certain things CxDs can’t do on their own. For instance, CxDs often have sufficient privileges to edit a bot’s text, but need a sysadmin to load the changes into the bot builder. 

Business analyst/data scientist: Once your bot has been up and running awhile, these are the folks who notice data points that require your attention, like bugs and drop-offs in key metrics.

Illustration depicting four groups of people a conversation designer works with: Domain expert, documentation, engineering, and marketing.

What You Do

You’ve got an idea of what the conversation design process looks like, and which stakeholders you’ll need to work with. But what will you actually be doing day to day? 

The actual number of hours each task requires will vary depending on the size of your project, team, and organization, so here’s a relative overview of the things you’ll be doing and where you’ll spend your time.

Illustration of two families, one all wearing size medium shirts and marked with a red X symbol, and the other wearing a variety of different size shirts, marked with a green check symbol.

The initial steps are:

  • Context setting, stakeholder alignment, and gathering requirements is a medium- to large-size task.
  • Research is a small- to medium-size task.
  • Creating your vision for the user experience is a medium- to large-size task.

The bulk of your work involves creating the architecture and copy that power your bot and revising/testing it ahead of launch.

  • Creating your information architecture is a large- to extra-large–size task.
  • Writing your discourse and copy is another large- to extra-large-size task.
  • Revisions, testing, and quality assurance (QA) is an extra-large–size task. This is your final chance to make sure everything works as it should before releasing your bot to the public.

What You Make

And what does all of that time and effort result in? Most conversation designers don’t actually code and deploy their bots—that falls to engineering and ops teams, as outlined. Broadly speaking, a conversation designer works on these things:

A Vision of What Conversational Experience Looks Like

Create the vision for how customers have conversations with your brand, whether through voice or text. Conversation designers in corporate environments often do this through a slide deck. Whatever format you choose, be sure to highlight what’s important: Voice and tone, inclusivity, relationship building, and helping users get what they need from every conversation. 

What You Think Your Bot Will Do

Here’s where you put strategy and tactics into practice. Outline what you expect your bot to do across the main use cases you’re solving for.

  1. Develop an information architecture flowchart that shows how your dialogs will fit together.
  2. Create a turn-taking structure between your bot and your user. Write out actual sample dialogs in your bot’s voice.
  3. Put that into a prototyper (Botsociety, Botmock, Voiceflow, Rasa) and output a video so others can get a taste of the real-time conversational experience. Alternatively, you can put the dialog into a spreadsheet and have the computer speak it aloud using text-to-speech.
  4. Create an annotated transcript of the turn-by-turn dialogs to help stakeholders understand your design process and strategy. The annotations should address points like, “This is why the copy is phrased this way here,” or, “Watch out for these trouble spots over here,” and so forth. The transcript can be delivered in a presentation deck, a spreadsheet, or whatever format is commonly used in your organization.

In addition to your sample dialogs, prototype or video demo, and annotated transcript, you’ll also want to create an information architecture for your bot. This is a flow chart that details the decision-making process your bot goes through—including what dialogs it speaks—depending on what the user says and does.

What Makes a Conversation Great?

We’ve taken a thorough look at what a conversation designer does, and who they do it with. But how do you make your conversations come out well? The next unit dives into strategies for conversation design.

Keep learning for
free!
Sign up for an account to continue.
What’s in it for you?
  • Get personalized recommendations for your career goals
  • Practice your skills with hands-on challenges and quizzes
  • Track and share your progress with employers
  • Connect to mentorship and career opportunities