Gracias por tu visita. Esta página solo está disponible en inglés.

Be Available

Build a horizontal, voice-first design that keeps all options open for customers. Don’t fall into the nested, vertical menu patterns that are used in GUI designs. Your skill should be able to handle any number of ways to trigger an action or intent.

Patterns and use cases

Create an effective invocation name

Customers open custom skills with a particular invocation name. This command needs to be specific to your skill and memorable. Often, customers combine the invocation name with an action, command, or question so your invocation name should be unique but not too long. For example, let’s say we have a skill named Daily Horoscopes. A customer could invoke and question at the same time with the phrase, “Alexa, ask Daily Horoscopes to give me a reading for Aquarius.” This phrase wakes Alexa up, opens the skill, and asks for a specific horoscope reading.

Design for response time limits

When Alexa is waiting for a customer to respond there is a limited time (eight seconds) in which the customer can respond before the service looks to either re-prompt the customer to see if they need assistance, or end the session. If your skill re-prompts the customer, it waits another eight seconds for a response before ending the session.

As with natural conversation, if the Alexa service thinks a question asked is misunderstood or confusing, re-prompts allow Alexa to clarify and reformulate a question to get the answer Alexa is seeking. Shorten a re-prompt for brevity when a customer is familiar enough with the context of a conversation that they won’t need the entire prompt again immediately. The key is that you provide enough information to guide the customer, understanding that you are essentially 8 seconds away from losing that connection if they don’t know how to answer. While re-prompts must be understandable, they provide an opportunity to expand on the initial request to get the conversation moving.

Do

Alexa: “Welcome back to Movie Talk. How can I help you today?”

User stays silent for 8 seconds triggering a re-prompt

Alexa: “You can hear movie times about new releases this week or you can listen to previews. Which would you like?”

Don't

Alexa: “Welcome back to Movie Talk. How can I help you today?”

User stays silent for eight seconds triggering a re-prompt

Alexa: “How can I help you today?”

Create a simple set of options

Avoid tasks involving information that the customer is not readily familiar with, or complex data, such as long alphanumeric passwords.

Do

Alexa: “I can tell you a story, recite a rhyme, or sing a song. Which would you like?”

Don't

Alexa: “I can tell a story, recite poetry, sing a song, tell jokes, tell you a rhyme, read the-”

User: “Alexa, stop.”

Create well defined, concrete tasks

Avoid correcting one word or part of sentence that was misrecognized, or correcting errors or ambiguity from prior voice inputs.

In this example, the customer is watching Prime Video and is asking Alexa to pause what they're watching.

Do

User: “Alexa, pause.”

Alexa automatically pauses the video with only a sound cue to announce she heard the request.

Don't

User: “Alexa, pause.”

Alexa: “You sure you want to pause?”

Video still plays in the background.

User: “Yes! Pause it! Pause it now!”

Alexa: “OK pausing your video “Man in the High Castle, Season 1, Episode 1 The New - ”

User: “Alexa, stop!”

Avoid ambiguous and complex tasks

Tasks such as searches should take any information the customer provides and provide contextually appropriate and short prompts to fill in any details that Alexa might need to complete the task.

Do

User: “Alexa, ask A to Z Restaurants to order me some pepperoni and olive pizzas.”

Alexa: “OK. That’s three medium pepperoni and olive pizzas. Would you like anything else?”

User: “No.”

Alexa: “Order placed! Your pizzas should arrive in 30 minutes.”

Don't

User: “Alexa, ask A to Z Restaurants to order me some pepperoni and olive pizzas.”

Alexa: “Would you like a medium or large pepperoni and olive pizza?”

User: “Three medium pizzas.”

Alexa: “Pizza Pro has two large pizzas on special right now. Want to start an order?”

User: “OK... But I want two pepperoni and olive pizzas.”

Alexa: “OK. You want two pizzas. What would you like on the first pizza?”

User: “I WANT PEPPERONI AND OLIVES ON BOTH OF THEM!!”

Alexa: “Pizza Pro doesn’t have any pizzas with topping ‘both of them’. What would you like on your first pizza?”

User: “Forget it- I’m not hungry anymore.”

Provide definitive choices

Organize your responses and prompts so that the customer has a clear choice to make. Open-ended questions can confuse the customer or cause them to answer in ways that you’re not expecting or supporting. For example, asking “What would you like?” is too open-ended. Even an either-or type of question, such as “Would you like Brie or Gouda?” opens up a likely response of “Yes.” Instead provide a brief, simple summary of available options first followed by a question prompting the customer to take action.

Do

Alexa: “We have Brie or Gouda. Which would you like?”

Don't

Alexa: “Would you like Brie or Gouda?”

Find the top three best matches to the input

Avoid information or data that needs further review or processing for understanding or follow up when the end goal is ambiguous.

Do

User: “Alexa, ask My Cookbook what I made for dinner last week. ”

Alexa: “Last week you made seven recipes. The most recent was Baked Salmon. Would you like to make that?”

User: “Sure.”

Alexa: “OK. Let’s get cooking. Baked Salmon for Two…”

Don't

User: “Alexa, ask My Cookbook what I made for dinner last week. ”

Alexa: “In the past, you’ve made Baked Salmon, Spaghetti and Meatballs, Turkey Meatloaf, Porkchops with Applesa-”

User: “Alexa, stop.”

Lists and engagement

Announce items in a list

For articles or lengthy information read aloud, avoid showing all items without first indicating how much information there is. Also, avoid search and browse tasks that involve large catalogs that consistently return too many results for the customer to explore via voice.

Do

User: “Alexa, ask My Trip Planner how many items are left on my packing list.”

Alexa: “You have 8 items left to pack: Flip flops, bathing suit, beach towel. Do you want hear more?

Don't

User: “Alexa, ask My Trip Planner how many items are left on my packing list?”

Alexa: “You still need to pack flip flops, bathing suit, beach towel, sunscreen, sunglasses...”

User: “Alexa, stop.”

Make lists brief

Have Alexa read the essential content in each list item; for example, always read titles, and only read secondary text if critical to the voice response. Generally, it shouldn’t take more than twenty seconds to read the first few items in the list.

Start with reading between two and five items, and adjust based on the following:

  • How familiar the customer is with the items in the list.
  • How long and voice-friendly the item names are.
  • The total number of elements read and displayed per item.

Do

Alexa: “Here are cheeses that you may like based on your flavor profile: Cheddar, Gouda, and Havarti.”

Don't

Alexa: “The cheeses you may like are Cheddar, Gouda, Jarlsberg, Porter Cheddar, St. Agur Blue Cheese, Gorgonzola, Brie, Gruyere, Sharp Cheddar, and Reggiano Parmesan.”

Ensure effective pacing with lists

Have Alexa say something to introduce the list, for example “Here are the popular quick meals,” and have her pause briefly between items in the list. Pauses and pacing help the listener distinguish where one list item ends and the next begins, for example:

  • Specify a comma plus a 350-ms pause using SSML after each item instead of a period or question mark. This makes the final item sound more similar to other items in the list.
  • Avoid adding an additional pause to list introductions that end with a period or question mark.
  • For lengthy list items or those that require the customer to think more deeply, consider replacing the 350-ms pause with a 400-ms pause.
  • Always test the experience by listening and then adjusting until it sounds right.

Prompt customers for Alexa to say more list items

When Alexa reads only a few of the possible items in a list, provide a way for the customer to tell Alexa to read more. When you know that your customers are interested in more than the first few items, have Alexa prompt the customer with a question like “Would you like to hear more facts about Seattle?”

Engage customers with questions

Remember, the customer always starts the conversation with Alexa, directing her with the type of skill they want to use. Once they do, you'll need to engage the customer to help Alexa determine how to assist the customer in the skill experience. Giving the customer a question is a natural cue to speak, prompting for customer input. Either lead with the question on its own or ask it at the end of your prompt so the customer will know to immediately respond. Asking a rhetorical question, or a question in the middle of a prompt, may cause the customer to begin answering before the mic is opened or before the prompt is done playing, which will result in a recognition error. Customers tend to answer questions immediately when asked.

Prompt with guidance for the customer

Generally, end with a question before having the customer respond. The question provides a cue to begin speaking and coaches the customer on what to say next. End the prompt right after the question so that customers don’t try to answer while Alexa is speaking. Be specific, but be ready for the customer to answer in a different way or to over answer.

Do

Alexa: “Let’s plan this trip. What city would you like to visit?”

Don't

Alexa: “Let’s plan this trip. I can help you once I know where you’re going and what you want to do there. Where would you like to go? You’ve been to Portland recently.”

Avoid unnecessary yes/no confirmations

If the prompt choices are yes or no, simply ask the customer whether they want to do something. do not prompt them with an additional “yes or no?”

Do

Alexa: “Do you want to keep shopping?”

Don't

Alexa: “Do you want to keep shopping, yes or no?”

Complete the task and end the skill session

Once your skill has provided the information that the customer has asked for, it's best to end the skill session. While you might feel like the customer might be losing out on discovering new features, using open ended questions to prompt customers is frustrating and lowers engagement over time. An open-ended question such as, "Is there anything else I can help you with?" puts high cognitive load on the customer to remember what your skill does. The frustration for the customer only compounds itself in noisy situations or when the customer has to explicitly end the session because the skill keeps prompting them to continue.

Do

User: “I would like Brie cheese.”

Alexa: “Great choice! Brie is best served at room temperature and can be paired with crackers and fruit, such as grapes.”

Don't

User: “I would like Brie cheese.”

Alexa: “Brie is a delicious choice. What else can I help you with?”

Back to top