During the September 27 edition of Office Hours, developers from around the world join Alexa chief technical evangelist Jeff Blankenburg. Blankenburg answers questions about Alexa, voice interfaces, and ambient computing.
Blankenburg was joined by people from the Alexa Skill Insights team to answer your questions. The Alexa Skills Insights Team gives personalized, data-driven recommendations to help developers take their skills to the next level. They joined Jeff Blankenburg in a recent edition of Alexa Developer Office Hours to share some of what they’ve learned through reviewing and elevating hundreds of skills.
From high-level guidance on best practices right down to the details that boost engagement, let’s dive into some of the key points from this session.
1. How do I submit my skill to the Skills Insights Team?
Guidance from the Skills Insights Team pays dividends; one skill builder even reported a 250% monthly customer base increase by using their recommendations.
So how can you take the first step to uplevel your skill?
Start by submitting a published skill—including the skill name and skill ID—to alexa-skills-insight@amazon.com. After that, the team analyzes it using a customer-centric approach, with criteria encompassing everything from the voice design itself to the accuracy of the skill’s description and categorization.
Once the Skills Insights Team has reviewed the skill, you’ll receive an e-mail containing unique recommendations and links to all relevant resources. But it doesn’t stop there. You have an opportunity to discuss the recommendations with the team via their office hours, conducted by the developers on a 1:1 basis. Finally, the Skills Insights Team sends a subsequent round of tips after the initial updates have been input.
Get the latest updates with the Insights Team’s Slack community at alexa.design/slack, or find their recommendations directly on the Skill Quality Coach (SQC) dashboard via the Analytics tab of your published skill.
2. What do high-scoring, engaging skills have in common?
According to the Skills Insights Team, there are two questions that help predict a skill’s success early on.
1. Do you have a unique value proposition?
2. Do you have fresh content to deliver on it?
As Omkar Phatak of the Skills Insights team described, conducting preliminary market research to understand what customers seek out is a crucial—yet sometimes overlooked—step. “Once you’ve done that,” he said, “half the battle is already over.”
Next comes achieving the skill functionality that drives retention. While there’s no one-size-fits-all model for a high-quality skill, Phatak notes that developers should keep iterating to improve accuracy and handle the most customer utterances possible. The ultimate goal is to develop an interaction model characterized by a robust, responsive voice interface.
There’s one more thing that can push a skill from “good” to “great”: personalization.
The team encourages developers to bake in a way to remember a user’s first session, then lead off the next session with a custom greeting message. This promotes retention by allowing customers the chance to build upon their progress, rather than starting from square one with each return visit.
3. Does FallBack Intent prevent me from processing wrong answers?
One viewer wondered whether Amazon’s Fallback Intent would prohibit them from processing false answers. As Jeff described, the answer is “yes and no.” Here’s why.
Consider a trivia-style game. In this situation, there’s one correct answer and a countless supply of false responses. If you want to catch and filter these wrong answers, one workaround lies in utilizing dynamic entities.
Dynamic entities allow you to replace the slot called ‘answer’ with only the correct response. Then, when an answer is presented, you can use a search query slot that catches everything—which means all user responses are captured and compared to the entity resolution in the ‘answer’ slot.
To see Jeff’s full interaction model in action, check out the GitHub page for TKO trivia.
4. Where should I begin with APL?
There are now more Alexa Presentation Language (APL) features than ever before, allowing you to build multimodal skills efficiently. APL is a proven way to boost retention and engagement by adding visual components to your skill, but unlocking its true capabilities doesn’t happen overnight.
In Office Hours, Jeff noted that “When you add APL, you double the size of the knowledge base required to develop. It’s a powerful tool, but it’s a lot to learn.” Luckily, there’s also a wealth of resources to help you get a head start.
Phatak of the Skills Insights Team advises beginning the APL journey through the Alexa Learning Lab, Amazon’s custom-made learning path with real-time feedback. The inaugural course centers on APL, taking you on a comprehensive deep dive from the basics all the way to advanced features.
By gaining experience writing and validating code all within the course, developers can leverage APL to open another dimension for voice skills—something the Skills Insights Team often recommends to attract users. For more inspiration, you can view live examples of skills incorporating APL like Ambient Visions and Kindspace.
5. I want to add instrument sounds to my skill. Is there a simple way to do it?
To create a memorable ambient experience, realistic soundscapes are essential.
But before you can include the perfect audio file to bring your skill to life, you need to ensure it’s licensed for commercial use and royalty-free.
For specific instrument sounds, there are online repositories like soundrangers.com that offer files for purchase and legal use. After finding and downloading the file, you’d then need to host it using a service such as an S3 bucket.
Another great place to begin is with the Alexa Skills Kit Sound Library. The Sound Library contains 3,018 unique sound effects with no need for file conversions or third-party hosting. There’s even an option to do it yourself by crafting personalized sounds and music with skills like Music Studio.
Ready to create unique voice experiences for the future of ambient intelligence? Explore these tips and more in the full Alexa Developer Office Hours replay, and don’t forget to tune in live to the weekly sessions on Tuesdays at 9 am PST.