Having been involved in digital technology my entire career, I was a little late exploring voice apps, but published my first Alexa skill in summer 2019.
One year ago I wrote about that process in a LinkedIn article titled “Building Alexa Skills for Kids - Advice, Stats, and Toilet Humour”, and here I will share some more tips gathered in the intervening period.
1) Target parents, not children
Organic discovery of 3rd party skills, especially on devices without a screen, is so small it’s not worth worrying about it. On skill exit Amazon will sometimes offer the user a “you might also like XYZ skill” message, but it’s clunky and I suspect does not get a high conversion rate.
Much more likely is for a parent to be searching via Google or the Alexa app for potential skills to be used for kids/family entertainment or education. So make sure to target for parents, and concentrate on the following:
- A skill icon that stands out and conveys quality (there are lots of poor “template” skills that clutter up the skill store)
- A detailed enough description to explain what the skill does, but not an essay that will put them off reading it
- Generally convincing the parent that the skill will either turn the child into a well-rounded genius, or keep them quiet for 10 minutes :)
2) Be funny
When writing stories, games, or information based skills for kids there are some variables, like the pacing or depth of content, that should ideally be adjusted based on the target age group of the users.
One thing that works fairly consistently is humour, and it is incredibly powerful for retention and advocacy.
When talking to users of my first skill Silly Stories, a frequent message was that the zany responses to their pre-story selections were as popular as the stories themselves. For example when asked to pick a type of drink, you would get a random response such as:
“Someone told me that the Queen of England drinks [drink] with her breakfast”
Or
“I once saw someone slip over in a massive puddle of [drink]”
3) Quality > Variety
Don’t kill yourself trying to pack your skill full of content up front.
Very young children in particular love familiarity, and will happily replay the same stories, jokes, or even quiz questions if the overall production value is good and enjoyable to listen to.
Silly Stories only ever had a maximum of 5 different stories available, but ~30% of users have played through 10+ stories total, with 5% in the 50+ range.
It is always possible to drip feed more content via CMS or in subsequent skill updates. It is more important to get real users using the skill, and hopefully providing valuable feedback via ratings/reviews to guide subsequent development.
4) Be careful with content
Amazon are required to put skills aimed at children through a more stringent certification process than for a standard general purpose skill.
Frustratingly there can also be a lack of consistency, and I have failed certification multiple times for changes that were entirely unrelated to the problem flagged. Often the cited issue had been live for many months already, and even without altering it could be let through in a follow-up submission, presumably by a different reviewer.
Below are a few of the reasons my kids’ skills have failed certification in the past:
- Mention of a pub (the context being a place to eat. It’s common in the UK for a family to eat together in a pub for lunch on a Sunday)
- Attribution link to a sound effects website in skill store description (considered as “advertising” and was required to be removed completely)
- References to violence and mature content(!) (annoyingly the rejection never specifies which bit of content is the issue here, but it’s one of the erroneous ones that tends to just be ok the next time submitted)
- Collection of personal information (there was a request for a character name, and this was considered as a risk of the user saying their own name)
- Inclusion of user generated content (The skill repeated responses of the active user, but never anything from a different user)
- Filtering out of profanity inputs from users (expectation to detect and reject inappropriate speech from the user)
5) Prepare to wait
Even 12–18 months ago when I was publishing my first Alexa skills for kids, the turnaround time for certification was multiple days or a week. However this has stretched recently to 1 month+ making it even tougher if failing for seemingly minor or confusing reasons and having to rejoin the queue.
I think this is partly due to the global pandemic impact, but also due to the upcoming release of the redesigned Echo Dot Kids. These devices are designed to be locked to child friendly skills only, and so the certification process to guarantee the quality level has caused a backlog of review activities.
Fingers crossed it will get back to more sensible timelines soon to re-enable developers of child targeted Alexa skills.
