This note is important for you to understand if you want to compare the stats from your August 2011 pdf reports to the stats from this month’s pdf reports. As you know our app download numbers are estimates and our our estimation model changes every month.
I. How We Constantly Improve Our App Stats
There are a couple of sources for our improvements:
1. Our models are based on raw download numbers which we get from publishers. Each month we gather more of this data. Subsequently we apply these models also to past data sets in the hope of gaining more accuracy.
2. Apple, Google and Microsoft publish their global download numbers. We always take those into account – our models, to a certain degree, follow global download numbers. Those are not published frequently and we estimate those global values for the following months. When new data is released we correct the estimations.
3. From time to time we simply try new things. We may invent better methods or we get insightful comments from the community.
II. What Does This Mean For You ?
Not much, overall – it only has significance if you take pdf reports from previous months into account.
You can directly compare download percentages in different categories or in countries.
You can compare apps counts and publisher counts.
However, you should not directly compare download numbers.
A proper comparison needs a recalculation of all past reports with the newest download model. We do not provide this for our free reports.
We publish, however, correction factors that you can apply to previous month’s download numbers to compare them with current month.
III. Correlation Factors August 2011 Reports & September 2011 Reports
As Apple published their global download numbers we changed the estimate download model for iOS. To compare August 2011 to September 2011 please multiply
- monthly downloads in the August 2011 report by 0.92 (we overestimated August 2011 downloads by 8%)
- total downloads in August 2011 report 0.96 (we overestimated total downloads at the end of August 2011 by 4%)
Previously Apple had announced their global app download stats in January and July, so we relatively content with this deviation.
In recent weeks we received a lot of new real download numbers for apps with downloads higher than 250,000. This changed our Android model.
To compare August 2011 to September 2011 please multiply
- monthly free downloads in August 2011 report by 0.88 (we overestimated Aug 2011 free downloads by 12%)
- total free downloads in August 2011 report 1.05 (we underestimated total free downloads at the end of Aug 2011 by 5%)
- all paid download numbers are without change and can be compared directly
In the coming weeks we will provide you with easier ways to submit your app stats so to improve our model.