At the request of Alexa’s VP, I delivered guidelines for how to create for a great proactive Alexa voice experience for an audience of product managers using a self-service tool to deliver proactive messages with Alexa. I also wrote a list of “guardrails” to use to avoid experiences that would feel like advertising, or feel pushy, aggressive, or repetitive to users.

Feature discovery is very difficult with a voice-forward interface, and especially a voice-only one. PMs are under pressure to deliver value for the business. They wanted to expose their features to more users and increasing engagement. With the development of a content-injection framework, they were able to do this.

Guidelines are no good if no one uses them! While delivering a document and socializing was my primary task, my ultimate goal was to earn trust and kindle interest with PM leaders. This tension between ownership and moving fast, versus seeking guidance and getting feedback – including potentially conflicing interests – made this one of the more challenging projects I’ve worked on.

I also had the challenge of selling the idea that guardrails are a good thing. PMs are trusted to represent the business and know best, and to gather data proving what works. Understanding their pressure to deliver, iterate quickly, I focused my approach on raising awareness in the tech team, meeting with every NLU domain owner, and most feature owners.

I also worked closely with the feature developer to recommend additions to the built-in guardrails, and gathered lots of feedback from leadership and product on my document to iterate toward the final version.

Finally, I built a wiki page using examples of using real Alexa TTS to give examples of how two experiences might clash and be annoying or incoherent to users, and how a PM might target an appropriate proactive experience. I presented this to the entire Alexa design team at our yearly all-hands, and also to the Alexa and Devices VPs and directors.