What I Learned From 5000 Reviews of Top 5 Alexa Skills

Amazon App Store screenshot

There are 16,000 Alexa skills with 0 reviews. They are Zombies. People never get back to them.

The way to avoid it is to precisely understand the use case for Alexa skill you’re building. To come up with some ideas, I went through top Alexa skills and read all user reviews, figuring out what makes them stand out. Here’s what I learned.


Design Alexa skills for repetitive experience

4 out of 5 top Alexa skills are designed like that.

What I mean by that is the skill gives you a little bit of value every day and updates that value regularly. It makes you stick to the skill to discover more & more content.

Jeopardy Alexa Skill page in the Amazon App Store.

Jeopardy skill is a perfect example of this. You can play only 6 clues every day, that’s it.

Every day the set of questions changes and it creates a fear of missing out.

That’s why people are constantly checking Jeopardy skill on their Alexa, they don’t want to lose those questions because they probably not gonna hear them again.

“Each morning I look forward to playing Jeopardy. When I miss a day the days doesn’t feel complete.”
“I’ve used the Jeopardy skill almost daily for about eight months with no issues until yesterday, when it just kept stopping during the 2nd question. Glad you were able to resolve the bug quickly!”

Use SSML to make Alexa’s voice more understandable

Here’s a #1 complain about every Alexa skill:

“She speaks WAY too quickly sometimes and I would like just 1–2 more seconds added to the timer. I miss about 25% of my answers because I am opening my mouth as time runs out.”

The way to avoid it is to use SSML when creating the content of the skill.

SSML is a markup language for text-to-speech systems. You can use it to set pauses in the conversation, increase/decrease prosody, and transform Alexa’s voice as you want.

Optimize for the use case

Based on 5000 reviews, here’s a list of most popular use cases that describes when & why people use Alexa skills:

Jeopardy:

  • Mornings — to get the brain moving before the workday, like a morning ritual
  • Evenings — family’s evening routine
  • Before going to bed — bedtime tradition
  • Before/after dinner — to spend waiting time, a great activity for the entire family while dinner is cooking or you’re just sitting around
  • After supper — to challenge the brain
  • While doing dishes — to unwind during the process

Fox News:

  • Mornings — to start a day with while making coffee
  • Evenings — to catch up all the top headlines for the day

Song Quiz:

  • Family time — to engage all family members in one activity, memory workout
  • Bedtime — to get a quick respite after the day
  • Mornings — to have fun before starting your day
  • Before/after dinner — to spend waiting time
  • While cleaning the house — “You want to clean? I hate cleaning my house. But if i play the song quiz at the same time I will be cleaning for 5 hrs!”
  • Before going to bed — great family activity, people usually play in pairs
People are using Alexa skills mostly to have fun with their family or to unwind while doing some unpleasant job (dishes, cleaning, waiting for something).

What optimization means is that you have to figure out the right use case why people are going to use your skill.

You can use any framework for that (I like JTBD the most, thanks Intercom). When you understand the use case, it becomes much easier to design the right experience and make important decisions like:

  • How long should one session be?
  • Do I need to make it sound funnier or more official?
  • Do I need to bring multiple players into the skill?

Don’t forget about timing, people love skills that don’t take a lot of time to interact with.

We are playing the game of attention here and you should care about user’s time.

Deliver useful and educational content through your skill

Most people like content that helps them feel better. And it doesn’t have to be a game, you can provide useful tips & tricks about your products, explain different how-to’s, provide them with the information they missed and much more.

“My husband and I start our morning with this skill. It’s a fun way to start the day and get those neurons firing.”

Choose premium features based on value

In Sleep and Relaxation Sounds example, people buy premium version for more sounds, higher quality, and sound combinations.

Sleep and Relaxation Sounds skill page in the Amazon App Store.

Here’s what they think about it:

“This skill is so useful to me that I gladly ponied up the $19.99 to enable the premium sounds and more importantly to support the skill developer. With premium, you can combine two sounds, and Tibetan Bowls and Rain is my favorite combo so far.”

That’s how you should think about the product. If the paid version is good enough to make people pay for it, they will.

But the biggest value proposition of this skill is that you don’t have to own a sleeping machine anymore.

“So I almost purchased an alarm for my son and a nature sound machine for my husband and I. I had one Echo, so I just moved it into our room and bought another for my son’s room and saved space and $20!! And I have way more sounds in the Echo than some best reviewed sound machines!! A++++”

Sleeping machine is a perfect example of how Amazon is disrupting another industry by adding a skill layer to Echo device.

You can think the same way and figure out the right use case where Alexa skill can replace the existing device/service/app that people pay for.


Key takeaways:

  • Choose the right use case with repetitive experience
  • Add new content to your skill on a daily/weekly basis to make people come back
  • Use SSML to set pauses, increase/decrease prosody and make Alexa’s voice more human
  • Optimize for the use case, try to understand the job the user is hiring your skill for (JTBD framework)
  • Deliver useful and educational content (tips & tricks about your product, how-to’s, provide them with information they missed)
  • Figure out the use case where Alexa skill can replace existing device/service/app people are already paying for

I’m excited to see where Alexa skills will take us and how we are going to communicate in the future while solving discoverability issues at the same time.


Thanks for reading! 🙌

Let’s keep the conversation going, you can connect with me on Facebook or join our community for Alexa skills developers.