Conversational components Questions

Questions

One of the most effective ways to get the user to continue the conversation (e.g., make a choice) is to ask a question. When the call to action isn’t clear, the user won’t know when, or how, to respond.

Think about what users could say Expand and collapse content An arrow that points down when collapsed and points up when expanded.

Before asking a question, think about what responses you can reasonably support. Don’t ask the user a question if you’re not prepared to handle their answer.

That said, don’t be afraid to ask the user a question—it doesn’t mean that you have to support every response imaginable. The way a question is phrased sets the user’s expectations for what they can say. This phrasing can range from open-ended, or wide-focus questions, to close-ended, or narrow-focus questions.

The magic and art of good conversation design is that users feel like they’re in control and that they can say anything at anytime, but in reality, the dialog directs them along pre-scripted paths.

Wide- to narrow-focus questions Expand and collapse content An arrow that points down when collapsed and points up when expanded.

Questions fall on a continuum from wide- to narrow-focus, based on the range of responses they elicit.

Image of a chart with examples of how a question can gradually go from wide to narrow. Wide focus question: What kind of bouquet would you like? The user can respond with a wide range of queries. Less wide focus question: What kind of flowers would you like in your bouquet? The user has a limited range of options, usually on a single topic. More narrow focus question: What color carnations do you want? The user can choose from a single category (for example, color). Narrow focus question: Are you ready to place your order? The user can choose from just a few options, such as yes or no.

When designing a question, think about where it should fall on the continuum from wide to narrow focus. Consider the pros and cons in the table below.

Pros

Cons

Wide-focus question

Best for questions about domains that are familiar to the user and therefore are easy to answer.

  • Encourages the user to respond naturally and in their own words
  • Users feel in control
  • Provides best insight into what users want
  • Can put users on the spot and cause a “deer in the headlights” moment
  • Can be difficult for users to anticipate what answers are supported
  • May set expectations too high and overpromise

Narrow-focus question

Best for questions about complex or unfamiliar domains, or when options are limited or unclear.

  • Makes it clear what the user can say/do by establishing boundaries/limits
  • Easy for users to answer
  • Minimizes confusion
  • Can feel limiting to users
  • Can feel robotic and tedious (like many automated phone systems)
  • Questions may be longer

Critical use cases for narrow-focus questions Expand and collapse content An arrow that points down when collapsed and points up when expanded.

There are a few specific use cases in which narrow-focus questions are critical.

Error handling

The first time asking a question, your persona should present only the details that are required to proceed.

When there’s an error, narrow the focus of the question to provide more support and get the user back on track.

See errors for details.

Persona: Sekai (fake shoe store) Spoken prompt: Which part did I get wrong? User input: (silence) Spoken prompt: So which part did you want to change: the color, the size, or the payment method?

Do.

Provide extra help when there’s an error.

Persona: Sekai (fake shoe store) Spoken prompt: Which part did I get wrong? The color, size, or the payment method? User input: (silence) Spoken prompt: So which part did you want to change: the color, the size, or the payment method?

Don’t.

Don’t slow all users down with details that most won’t need to answer the question. Remember, irrelevant speech is time consuming.

Disambiguation

Language is filled with ambiguity, though most of the time it is resolved through context. When context is not enough, it’s okay to ask the user for more information.

Persona: Miso flowers (fake flower shop) User input: Send mom a few roses. Spoken prompt: Alright. For color choices, I recommend either pink roses to convey elegance and appreciation, or yellow roses to convey friendship and joy. So, which color do you want?

Do.

To resolve ambiguity, ask the user a narrow-focus question.

Persona: Miso flowers (fake flower shop) User input: Send mom a few roses. Spoken prompt: Okay, I’ve added a dozen pink roses to your cart.

Don’t.

The user’s request is ambiguous. This response would’ve been appropriate if the user has ordered pink roses for their mom in the past.

Explicit confirmations

Though they’re uncommon, explicit confirmation questions are necessary in these cases:

  • When the cost of misunderstanding the user is high (e.g., names, addresses, text to be shared on the user’s behalf)
  • Prior to performing an action that would be difficult to undo (e.g., deleting user data, completing a transaction)

See confirmations for details.

Persona: Sekai (fake shoe store) User input: Delete everything on my wishlist. Spoken prompt: There are 5 items on your wishlist. Do you want to hear them first, or should I go ahead and delete them?

Do.

Before deleting user data, use narrow-focus questions to confirm.

Persona: Sekai (fake shoe store) User input: Delete everything on my wishlist. Spoken prompt: Okay, they’re deleted. Do you need anything else?

Don’t.

It would be difficult to recover from an error here.

Learn from what users said Expand and collapse content An arrow that points down when collapsed and points up when expanded.

The insights gleaned from user data are invaluable, shedding light on how to better phrase your questions. Analytics and monitoring tools can help you learn how your questions are performing in the wild. Review all the ways users have answered a particular question, looking for those your Action failed to handle appropriately. Also look for signs of user confusion—for example, perhaps users are not responding (No Input), they take a long time to respond, or they’re hesitant, using filler words like um and ah.

There are 3 main approaches to handling unsupported responses:

  • Add new synonyms to the grammar so that they map onto existing functionality
  • Restrict the range of responses users can give by narrowing the focus of the question
  • Design a new conversational path to support the requested functionality

If you’re seeing a lot of unexpected responses from users, consider rewording the question to narrow its focus.

Persona: Miso flowers (fake flower shop) Spoken Prompt: Which color roses do you want: red or white? User input: White.

Do.

Rephrase questions to make the options clear.

Persona: Miso flowers (fake flower shop) Spoken Prompt: Do you want to use red or white roses? User input: Yes.

Don’t.

Typically, intonation and context will make it clear to the user that this is an either/or question, not a yes/no question. But if users are having trouble, consider rephrasing.