He currently manages all APIs and backend infrastructure for Sportsman Tracker, which offers mobile apps for fishing and hunting enthusiasts, called FishWise and HuntWise, respectively.
Tell us how you got into building Amazon Alexa skills?
I’ve always been a fan of voice in the car, and thought voice would be cool in the home. I waited because I thought the Echo was too expensive, but when the Dot came out and I was in to try it.
My first skill was a daily deal tracker, Bargain Buddy, because I love daily deals like Woot.com.
At first, my skill got rejected by Amazon, and I responded by adding a few other deal sites into the skill. It turns out that made the skill better, so the initial rejection wasn’t all that bad 🙂
When did you “crack the code” and get real traction with Alexa skills?
I was disappointed by Bargain Buddy’s adoption, but I understood why. Then I looked to see if Alexa could look up stock prices, and was surprised it couldn’t – so I built Opening Bell. It was useful and consumers adopted it, but still not at a large scale.
The big ‘aha’ moment was when I moved into a new apartment and I had an upstairs neighbor who was really, really noisy. I thought, “man, it would be sweet if Alexa could play ambient noises to help me sleep.”
I built Rain Sounds, my first skill with strong traction. It got consumer adoption through no advertising, purely organic, solely based on consumers searching for it because they wanted it. That adoption spurred my next two skills: Thunderstorm Sounds and Ocean Sounds.
One of the things I did right was to keep it simple. Rather than create one skill that had several different sounds, I created a separate skill for each sound. It is very important to let a user say “Alexa, open Ocean Sounds,” instead of having to traverse a menu.
Keep the consumer requests and responses short and sweet.
What have been your biggest challenges so far?
Managing and reacting to user feedback (reviews).
With Opening Bell, I gave the consumer a lot of options, which is hard to get perfect because there are 1000’s of obscure company names.
Despite reacting quickly to user feedback to teach Alexa how to better handle company names, the skill gathered a short string of 1-star ratings from users who expected nothing but perfection. These ratings quickly caused newcomers to turn away, leaving the skill in what I call “review purgatory.” Before, there was no easy way to respond to a review, and even when I did it was very rare that a consumer changed their review – even after I resolved the issue the user was experiencing.
You get one chance, and one bad experience can cause a 1-star review that scars your rankings forever.
Learning from this, I implemented VoiceLabs into Opening Bell to understand which company requests were not working, so I could try to prevent 1-star reviews as quickly as possible.
What’s your advice to people building voice apps?
- Keep the interactions short – ask basic questions and don’t get too wordy with your responses
- Scratch your own itch – your best work happens when you have a personal interest in the outcome
- Think about how the everyday consumer will try to find your Skill – a strong set of search keywords go a long way
- Give your Skill an obvious name – a user’s first impression of your Skill will only be based on the Skill’s name. If the name alone doesn’t tell them what your Skill does, then they probably won’t bother to tap into the detailed description.
- Test your invocation name… a lot – and assume that users haven’t completed Alexa’s voice training. If you can’t activate your Skill with your invocation name 99.9% of the time, you should probably consider a different invocation name.
- Fail gracefully – A friendly error message will be received better from users than a generic time-out message from Alexa.
- Gather and react to statistics – an invocation error alarm for your Lambda functions (if you’re using Lambda) can help you react quickly to increased failure rates. Using an analytics platform like VoiceLabs can help you understand how users are using your Skill to help you focus on what aspects need the most attention. On a big traffic day, I checked VoiceLabs every 20 minutes to uncover surprising results and actionable data.
- Be active in the developer community – there’s an (unofficial) Alexa Slack Organization with over 500 people registered and at least about two dozen folks actively listening on it at any given time (even a few Amazonians). Shout-out to Travis Teague, who organizes the group and has done a great job. We bounce ideas off each other, test each other’s Skills, answer technical questions, and partake in random banter on a regular basis.
How do you think about getting consumers to try your Alexa skill?
Think about how the everyday consumer will try to find your skill. Setting an obvious Invocation Name and common-language search keywords are super important when submitting your skill to Amazon.
For example, I called Rain Sounds “Rain Sounds,” not “Ambient Noise Sounds”. I put those words in after-the-fact, but focused on ‘consumer-first’ naming.
You only get one shot to set the Invocation Name with Amazon. However, you can change the Search Keywords and should fine-tune them as your skill evolves.
How have you gotten distribution? How have you gotten so many users?
By accident… it all comes down to scratching your own itch. On the first day of releasing Rain Sounds, with no promotion, I got about two dozen positive consumer reviews. Consumers were hungry for what I was building.
What’s next for you on your Voice journey?
I am going to double-down on my existing skills. I have received a lot of feedback and traction with Rain/Thunderstorm/Ocean Sounds, and I want to leverage that the best I can. Maintaining good ratings is key to a skill’s survival/growth.
What is your advice to Amazon?
Create a way for bad reviews to fade away or be marked as resolved. A small streak of 1-star reviews, because of a temporary or niche problem, can scare newcomers away from trying your skill and kill great skills.
To prevent trolling, perhaps make a review removal process part of the recertification process.
What is your advice to Google?
Create the “Dot” of Google Home, so consumer and developer adoption is an impulse-buy. $50 is a heck of a sweet spot.
I would love to expand my development to Google Home, but $130 is a bitter pill to swallow to develop on a platform that doesn’t have a clear monetization strategy for developers yet.
Who are other skill developers you think are innovating?
Jo Jaquinta has developed some interesting stuff. I really appreciate when a developer has a lot of attention to detail, and that radiates from his skills like Sub War and 21 Blackjack. They’re not best examples of short and simple voice experiences, but they’re definitely the work of someone who loves their craft.
Adam Marchick is the CEO of VoiceLabs, a company dedicated to evolving Voice Applications in the home, car and on the go. VoiceLabs’ initial product is an Amazon Voice Analytics service, which is now the most widely used Alexa service and serves as the foundation for data-driven growth products.
VoiceLabs works with great companies and developers as they adapt to the millions of consumers now using their voice to interact with services, providing engagement and monetization solutions in the emerging Voice ecosystem.
Nick Schwab is a serial entrepreneur. During his day he manages all APIs and backend infrastructure for Sportsman Tracker, and at night he is one of the top 10 independent Amazon Alexa skill developers in the world. A few of his Voice Application hits are Opening Bell, Rain Sounds, Ocean Sounds and Thunderstorm Sounds.