Conversational Actions will be deprecated on June 13, 2023. For more information, see Conversational Actions sunset.
Carousels are optimized for allowing users to select one of many items when those items are most easily differentiated by an image. Users can select an item by either saying its title or tapping it.

Example

Here’s an example of what a carousel looks like when all required and optional fields are completed.


Requirements

This visual component currently supports customization.

Field name Required? Restrictions/Customizations
Item image No
  • Choose from three different image aspect ratios: square, landscape, and portrait.
  • Default size depends on screen size and aspect ratio; any extra space will fill with bars.
  • Image source is a URL. If an image link is broken, then a placeholder image is used instead.
  • Alt text is required for accessibility.
  • Customizable image shape (angled or rounded corners).
Card background No
  • Customizable image or color.
Primary text Yes
  • Each item’s primary text must be unique (to support voice selection).
  • Plain text by default. Fixed font and size.
  • Max 2 lines recommended. Depending on surface, additional characters will be cut off.
Secondary text

Also called body or formatted text.

No
  • Plain text. Fixed font and size.
  • Max 2 lines recommended. Depending on surface, additional characters will be cut off.

Number of items

  • Maximum: 10
  • Minimum: 2

Consistency

All items in a carousel must include the same fields—e.g., if one item includes an image, then all items in the carousel must include images.

Interactivity

  • Swipe: Slide the carousel to reveal different cards.
  • Tap: When users tap an item, the item’s title is accepted as the user input, starting the next turn in the dialog.
  • Voice/Keyboard: Replying with the card title is the same as selecting that item.


Guidance

Carousels are mostly used for browsing and selecting among images.

Use carousels to help the user select from content that:

  • can be most meaningfully browsed via scanning imagery (e.g., movie posters, album art, recipes, clothing)
  • can be meaningfully blocked into rectangular chunks (e.g., tweets, news stories)

Users will be able to say the item’s title to select it, so make sure they’re easy to say, and uniquely identify each item.

Do.

Each item title should be as short as possible while staying distinct from the other items.

Don't.

Never use the same title for multiple items. And avoid titles that are very similar.

Consider including information about the following:
  • How many items are in the carousel (e.g., “There are 7 items on your wish list.”)
  • Why these items were chosen (e.g., “Here are our most popular bouquets.”)
  • Any selection criteria on for the items (e.g., “concerts this weekend”)
  • What order the items are in (e.g., “starting with the most recent order” if reverse chronological)

Do.

Let the user know why you’ve suggested these specific items.

Don't.

Don’t leave the user wondering why your Action is showing these specific items.

Ask a question to let the user know to take their turn. Include chips like “none of these” to let them indicate they don’t want any of the options.

Do.

Make it clear to the user that they need to select something from the carousel. Here, the chips allow them to choose “none of these” or to refine the results (for example, by only showing “patchwork sneakers”).

Don't.

Don’t simply show the user a carousel. Ask them a question in a way that makes it clear what happens if they choose an item.