Friday, October 31, 2008

Website traffic estimation: My excel forecasting model (Part 2)

Due to the number of inquiries after my first posting "Website traffic estimation: is there..." I've decided to share my forecasting excel model.

To download: http://www.scribd.com/doc/7668443/Traffic-Estimation-Model-1




IMPORTANT

1) Linear Function: y = 8E+09x + 49771
2) 2nd degree polinomic function: y = -1E+13x2 + 1E+10x - 3272,2
3) For the part of the curve that corresponds to websites with an Alexa traffic rank below 100.000, the function used is an average of the polinomic and linear functions

Two functions are used to calculate traffic. This could be upgraded, corrected and improved with more data points or sources of datapoints.

Please if you know any public source of reliable data of historic data on traffic, or have any comments email me at pedroneiraf@gmail.com

4) Cells with yellow background ARE NOT Alexa's 3 month average reach. This input was distorted, so before inputing the data directly from Alexa, it has to be corrected
5) The way in which this data is "corrected" is by a simple cross-multiplication aproximation to the nearest upper value in the list (the closest upper white cell).
6) This function will not work well for recently created websites.
7) This material is for non-commercial use only. Any usage of these functions or part of this excel sheet should quote the source.


The results I've got so far with this function have an average error of 7%, although it has to be said that in some cases, the error might be up to 40%. It has worked for me so far 80% of the time with an error of less than 15%, which is more than acceptable considdering the errors wound in other tools like the web Traffic Estimate. If you have a website and know your numbers, just compare and see the difference.

Please, this model is just a starting point, that needs to be upgraded, corrected and improved with all your comments.... Any suggestions ???

Copyright MBA Internet Marketing Manager


No comments: