alexa_device1.png

Voice Interface Design

Company: The Economist Group

Objective: Create a unique voice experience for Economist news based on the Espresso app

The main goal was to have the user ask for economist news and Alexa would read the articles to them in a format similar to the Espresso reading experience. As UX lead I was tasked with designing the voice interface for The Espresso skill and what that interaction would sound or feel like. (skill meaning an app you might use on the echo device)

Prototype

Research 

Competitive + Comparative Analysis

I began by looking at other news "skills" to see what they offered on Alexa and what they could do. I found virtually all of them only offered a flash briefing which is short form version of content (like an article) that gets served up to the user in either text-to-speech or audio. I thought we could do better. Moving away from news, I explored some other skills which gave tremendous insight into what Alexa was capable of. Particularly the "Wayne Investigation" which showed how navigation and user choice was utilized to deliver a story.

Key Takeaways

  • Cognitive load - limiting options to no more than 3 at a time to not overwhelm the user.

  • Intents - Avoiding 'yes' and 'no' responses, directing users to specify their choice helps Alexa understand.

  • Taxonomy - Understanding what terms users thought they could say to interact.

  • Utterances - affording for variations in user requests that deliver the same result.

  • Error handling - making sure the user has a way to recover if Alexa doesn't understand.

App Map and User Flows

I built an app map to understand the way the content was structured and any potential design constraints. As we began iterations of the flow we discussed with amazon what the best practices were in designing for voice. We started to strip away certain elements and tailor the experience to be more conversational. This lead to better error handling, smoother transitions in navigation, and clear distinctions of where the user was and what they could do next.

the final user flow is below.