4 ms·

This is not actually true -- I ran some analysis on review data and when controlling for downloads there isn't a strong correlation between ranking and # of rev

by diziet 13y ago

This is not actually true -- I ran some analysis on review data and when controlling for downloads there isn't a strong correlation between ranking and # of reviews. For example, https://sensortower.com/ios/us/funzio-inc/app/modern-war/468327549 https://sensortower.com/ios/us/funzio-inc/app/modern-war/468... Modern War had a spike on Feb 9th -- ranking stayed as predicted. Another example: Facebook https://sensortower.com/ios/us/facebook-inc/app/facebook/284882215 https://sensortower.com/ios/us/facebook-inc/app/facebook/284... on Jan 23 or Smule Dec 5th: https://sensortower.com/ios/us/smule/app/magic-piano-by-smule/421254504 https://sensortower.com/ios/us/smule/app/magic-piano-by-smul... A lot of apps can push user messages say via urban airship to prompt for reviews, and we see events like this all the time.

michaelone 13y ago
Interesting, do you have data on Flappy Bird itself? Apple might normalize their data for those kinds of anomalies like Modern War's blip, but your Magic Piano example does show a rise in ranking both of the times they had an upswing in reviews. Apple does seem to weight installs over time, so I wouldn't be surprised if they also weight reviews over time to catch those kinds of push notification strategies. Zach Williams's data showed something quite different than Modern War's: more of a snowballing after the initial climb, so maybe the consistent influx of reviews from the in-game button was a "normal" enough data point to get past Apple. Nobody obviously knows for sure all of the details of Apple's current algorithm.
slaven 13y ago
Also not true: Apple doesn't get "Engagement" data from the app, so that part of the ranking formula is wrong.
muzzamike 13y ago
Thanks, this is good to know. The other large component which gets thrown around is engagement. It seems as if only companies like flurry would be able to discern this sort of information. Anything on a large scale with an embedded sdk that reports engagement really.

by diziet 13y ago

michaelone 13y ago

Interesting, do you have data on Flappy Bird itself? Apple might normalize their data for those kinds of anomalies like Modern War's blip, but your Magic Piano example does show a rise in ranking both of the times they had an upswing in reviews. Apple does seem to weight installs over time, so I wouldn't be surprised if they also weight reviews over time to catch those kinds of push notification strategies. Zach Williams's data showed something quite different than Modern War's: more of a snowballing after the initial climb, so maybe the consistent influx of reviews from the in-game button was a "normal" enough data point to get past Apple. Nobody obviously knows for sure all of the details of Apple's current algorithm.

slaven 13y ago

Also not true: Apple doesn't get "Engagement" data from the app, so that part of the ranking formula is wrong.

muzzamike 13y ago

Thanks, this is good to know. The other large component which gets thrown around is engagement. It seems as if only companies like flurry would be able to discern this sort of information. Anything on a large scale with an embedded sdk that reports engagement really.