Conversational analytics solution Dashbot has a new feature that analyzes Alexa skill reviews. Voicebot conducted the first comprehensive analysis of Alexa skill reviews back in September and found that 62% had no reviews at all and 5% had only a single star rating. Dashbot’s more recent analysis also found that 62% still had no reviews and 16% had only one rating. It is tempting to say the 62% are the same, but the total Alexa skill count has risen by more than 4,000 since that time. So, we are seeing a consistent level of skills receive no ratings as the total skill count has risen 20%.
The big difference is that the percentage of skills with only one review tripled. That also means that the percentage of skills with more than one review actually declined during the period. Dashbot also found that the top 5% of skills in terms of ratings have nine or more. That total would reflect about 1,200 Alexa skills with reviews.
Ratings Tend to Bunch in the Middle
Amazon has a five-star rating system for Alexa skills. There is no zero. So a 1 is the lowest rating and a 5 is the highest. Dashbot found that 1 and 5 star ratings were likely to be solitary for skills. For example:
70.5% of one star reviews are from a single rating, and 65.6% of five star reviews are from one rating.
If you look at the Alexa skills with two or more reviews you see a fairly clear but broadly dispersed normal distribution with spikes at the limits.
Dashbot also narrowed this analysis to consider only the Alexa skills with nine or more reviews. This provides a more pronounced normal distribution, eliminates the spike at the low end of ratings and enhances the bulge for rating at or near the top.
This shift to the right for the ratings distribution makes sense. If you assume that the number of ratings is a function influenced by total users and that skills with low ratings will attract fewer users, then removing skills with few reviews will also disproportionately remove skills from your analysis with lower average reviews. Conversely, more popular skills are likely to have more reviews and higher ratings.
Do Alexa Skill Reviews Matter?
Some people suggest Alexa skill reviews do not matter. The hypothesis rests on the notion that few people use the Alexa skill store to browse and therefore are unlikely to ever see the ratings. Also, the voice experience does not lend itself naturally to the visually driven review process. However, for people that do use the skill store, there is a way to sort your search by skill rating. That invariably helps improve visibility. There have also been repeated comments in forums that Amazon helps promote only those skills with high average ratings. These platform promotions have reportedly led to eight times increase in daily skill usage.
There are many factors beyond those listed above to consider. Growing evidence suggests overall ratings and total ratings will make a difference. We just don’t know how much difference at this point. For those interested in tracking their Alexa skill ratings and comparing them to other skills, Dashbot’s new feature should be a welcome tool to hit the market.