How to Write for Alexa Skills

If you haven’t yet used Alexa Skills, don’t worry. It’s coming. Alexa Skills is the equivalent of an app on any other device. When you download a Skill, it adds to Alexa’s knowledge. And if you’re a content strategist or content creator, chances are good that you’re going to be creating a Skill in the near future.

But how do you write for Alexa, Siri, or Cortana? It’s surprisingly difficult. As I set out to create my first batch of content for a Skill, I assumed I could use content directly from my client’s website. I was wrong. Although the copy read well on the website, the content sounded stilted in audio.

Writing for Skills vs. Writing for Other Voice-Based UI

One thing to remember is that Alexa Skills is not a bot. A bot is based on AI. It has multiple pathways to help lead you in one direction or another. How does Skills work?

  • Where bots need to consider the flow of a conversation, Skills, on the other hand, are designed to answer direct questions.
  • If you are developing a bot, you need to consider the technology. Skills are built to work with Alexa; you have no control over the technology.
  • In addition, when you develop a bot, most people focus on how to create the voice. But Alexa has its own voice and tone. Skills has to fit into that voice; it would be odd for Alexa to lose consistency.

You might think it would be better if you could create your own branded Alexa. After all, Google is now offering British, Australian, and Canadian accents. You can choose a male voice (possibly removing some of the sexual harassment issues?), or even, purportedly, John Legend’s voiceBut Alexa needs her own voice. 

Here’s why.

Alexa has no Visual Context

Structured content has changed the way we create content. It forces you to consider context. And when it comes to voice-based user interfaces, the context is entirely auditory.

That means you have no:

  • Breadcrumbs
  • Navigation
  • Header
  • Sidebar content
  • Related elements

In other words, anything you want your user to know needs to be incorporated in your answer. The spoken words are the only content that exists. There is no additional context available!

Alexa has Limited Knowledge

Yes, voice technology has made extraordinary strides. But Siri, Alexa, and Cortana still have significant gaps. For that reason, they are designed to ask forgiveness. Many Skills are prefaced with statements that imply that the UI is still learning. For example, “I wondered about that too!” or “That’s a tough one, but I think I can help.”

That’s on purpose. When you see the UI as a peer, you are more forgiving of her errors. When Alexa makes a mistake, you are more likely to think “that’s ok, you’re still learning!” and less likely to get frustrated.

Few brands are going for uncertainty. Unless your brand is one of them, you probably want to keep Skills content in Alexa’s voice, and site content in your own.

Tips for Writing in Alexa Skills

Are you building a new conversational UI? Awesome! Ignore everything in this article. But if you’re writing for Alexa Skills, give this some thought:

  • Keep sentences short and sweet
  • Ask “do you want to hear more?” after the first sentence or two
  • Ask “do you want to know about [related information]” or “do you want an example?” after answering the question; this continues the conversation
  • Focus on Alexa’s voice, not your brand voice

Conversational UIs are great. But Alexa Skills isn’t one of them. You can help your customer and help your brand by developing a skill; just remember that the voice is not your own. You’re in Alexa’s world now!

2 comments Add yours
  1. Was the section “Alexa has no Visual Context” supposed to end with “The spoken words are the only *context* that exists”? It’s the only way that makes (interesting) sense to me.

Leave a Reply

Your email address will not be published. Required fields are marked *