Looking through our data, I noticed that several skills names appeared multiple times. This was not particularly surprising and duplicate entries are not unique to the Alexa marketplace: both the App Store and the Google Play have similar issues. However, when I took a closer look, I was surprised by the scale of skill name duplication for such a relatively small marketplace.
There are 1236 skill names shared by more than one published skills. In total, there are 4034 skills that share their name with another skill. 72% of these skills (2900) have 0 ratings (ratings are used as a proxy for user engagement). The majority of these skill names (64%) have one duplicate. However, some skills have over 15 duplicates!
Judging by these skills’ names, they are likely published as part of how-to guides or someone’s attempt at publishing Amazon Skill Blueprints. One of the Alexa ecosystem’s strength, and perhaps a weakness, is the ease with which you can publish a skill. Most of these duplicates probably aren’t malicious. They’re simply someone playing around with the platform and publishing their first skill.
The duplicates in the graph are not for particularly popular skills, except for Dad Jokes. One of the 35 Dad Jokes skills has 975 ratings! It’s notable that this iteration of Dad Jokes uses “my dad” as an invocation, instead of the “dad jokes”like the other 34 skills. Could uniqueness play an important role in discoverability?
There are two other curious cases where both duplicates are fairly successful: Repeat after me with 7303 ratings and 1128 ratings, and Pull my Finger 681 ratings and 610 ratings. I don’t understand how this can happen. I would expect the recommendation engine to favor one skill over the other, especially in the case of Pull my Finger. Strange.
How do you interpret this relatively high volume of duplicate skills? If there are now around 70 000 skills in the US marketplace, 4% of these are duplicates. Is it a sign of a healthy and mature marketplace or does it imply a problem with the ecosystem?