For designers and developers looking to build conversational Alexa skills, you have to think differently about how you design the experience. The hard-won design patterns we've found success with on other interfaces like the web and mobile haven’t always transferred seamlessly to voice. Voice experiences are best when they're adaptable, letting customers speak in their own words. They're personal, speaking directly to customers. They're available, allowing customers to set direction. And they're relatable, cooperating with customers to get something done. Each of these voice patterns boils down to carefully choosing what we say and how we say it.
In much the same way the layout, color, and animations on a screen can impact usability, wording can signal subtle meaning to users. After you've written a script for your Alexa skill, charted the edge cases, and mapped dialog flows, it's time to turn your attention to wording. These small adjustments to your skill wording can have major impact on how focused, natural, and simple the voice experience is for your customers.
I recently spoke with Alexa Senior Solutions Architect Justin Jeffress on this topic during Alexa Live, a free online conference for voice developers. During our session on How to Write Great Dialogs for Alexa Skills, Justin and I shared best practices for writing dialog that can elevate a voice experience from a transaction to an experience. In today’s post, I share a recap of what we shared during the session. You can also watch the full 45-minute session below.
In order to write great dialogs, you need to start with the right foundation. Conversations ebb and flow, and the responses depend on the previous answer to a question. When designing your dialog for a skill, it might be very tempting to use a flow chart (pictured at 2:47 in the video above). We’ve found that flow charts drastically limit the interactions customers can have with your skill.
The best skills are dynamic. They adjust to what they learn about the customer and flex based on the different ways a customer might interact with a skill. Instead of using flow charts to design your dialog, use situational design to build your foundation. Situational design is a voice-first method to design a voice user interface (VUI). Using situational design, you start with a simple dialog that helps you focus on the conversation and how customers interact with your skill. Learn more about situational design here.
When you’re getting to know another person, you get to know more and more about that person over time. Every time you see them, you don’t have to re-introduce yourself. You pick up where you left off and continue the conversation. Your dialog should adapt to the customer and their familiarity with your skill. For example, how long have they used your skill? Is this their first time? Is this the second time? How long have they been away from your skill? Has it been five seconds? Five days? Five months? If it's been five seconds since the last time that they used your skill, then you probably don't need onboard them again right away. But if it's been five months, maybe it's time to let them know that you’ve added a new feature instead of treating them like they've never used the skill before.
With screen-based experiences, the customer can look at a screen to get the information they need and know what to do next. With voice-first experiences, customers have to rely on your dialog. They have to listen, they have to process, and they have to respond. You need to make sure that you're writing things to be heard, not read. That means being brief in your responses so customers can digest what Alexa is saying. It also means pacing your dialog in a way that feels natural.
I always like to tell developers to “break the grammar rule” and don’t be afraid to use punctuation. Alexa respects commas and grammatical punctuation marks with pauses in her responses. If you use a comma in your dialog, she'll naturally pause. In a natural conversation, we frequently pause and take breaths. Putting commas in your dialog will help Alexa slow down so the customer can better digest the information. We cover several examples starting at 2:11 in the video.
In addition to the above, our video covers tips for:
Watch the full session to get all the best practices.