Any other reasons - I don't know - I have build a liner regression model to even estimate the payout estimate - and I clearly see that they have changed their payout structure. Feel free to PM me to know about this . ( Not sure if its okay to publish the structure - or even that , the model i estimate is anywhere close to correct )
Since I received a few requests to explain this part - here is the long explanation to this . Here's a preview of what I did .
If you are comfortable with linear regression and you want to do it yourselves , you can just go to your a-ads site ,
get the data in this format for existing data :
earnings = x1* impressions + X2*as_unit_clicks + X3*affiliate_clicks + Error Error = the error tolerance ( because of other parameters that are not used in the equation - like quality_of_site )
Note: if you donot have any affiliate_clicks, over a few months of data , this give pretty much accurate results .
Compare it with earnings from a ad-units before last month and then with one in this month. This will give you the two different equations, with two different payout schemes .
Note: To get your linear equation , you can use any tool ( python, R ) or even you can use any online linear regression calculating sites .
Happy to help anyone further - as usual, PM me.