These are great and I look forward to you iterating on this - but I fear your over-reliance on Alexa is producing semi-worthless results. I can go into all of the reasons why Alexa data is unreliable but I assume you already know (the big one is selection bias based on who has the toolbar installed).

If you combine Alexa data w/ data from Quantcast, Statcounter and others, it would probably be valuable. But even Alexa themselves would tell you their dataset has massive holes in it when trying to use it as the foundation for this kind of thing.