Share

As we built the App Score, the quality score for apps, and started to analyze the patterns of ratings and correlation with rankings we observe that a significant number of apps, including top ranked had a suspicious number of rating collected – often very positive (4 or 5 stars only or mostly)

Ratings are not interesting as a still data, they are interesting if you manage to deduct a pattern that imply a level of quality for example if over time your app consistently get high ratings or low ratings. But this is meaningful only if associated to a logical level of download activity or ranking pattern.

For example if your app is one day old and you collect thousands of positive reviews, when the 10 most downloaded/ranked apps in this store collected about 10% the same amount of ratings over the last 12 months, you can have reasonable suspicion on the meaning of this rating pattern. Ratings become meaningless and users need to know that they can’t rely on this information.

You can surely improve how ratings are collected by suggesting your users to rate your app. But we’re taking about massive patterns that are out of the ordinary practices. And those who use them, know what we’re talking about.

App score is now introducing rating corrections, that will apply to apps who have experienced suspicious rating activity.

This correction is starting to roll out today progressively and will reach all the app population in the coming weeks.

You may then observe that many apps will be affected in their overall score.

We’re constantly improving the App score. We’ll introduce more parameters soon to improve quality detection and make it simple for users to decide whether or not an app is worth downloading.