How accurate is Alexa for traffic measurement?

I think you are spot on in identifying Alexa as being a broad indicator that can warn you that something is afoot.   If someone asks me how far it is to the pub, the fact that my answer could be only a rough guess still makes it better than nothing.  Actually a lot better than nothing. 

Take our b2b site www.bobsguide.com whose alexa ranking has varied between 50,000 and 100,000.  That alone says the figures are dodgy.

But along comes Mr Client who says they would rather place their ads with http://www.btnsn.com/ .  It is very useful to me to know that their alexa ranking  is around a million. 

In the world of high traffic b2c sites you can rely on the figures in the media packs being pretty much OK.  Once you get to world of niche b2b sites things are different.  There are many sites from big important publishers who's alexa ranking is in seven figures. 

http://www.bankingtech.com/
http://www.risk.net/

Ads on these specialised sites are not sold on a CPM basis.  I remember having an interesting conversation with one site (who shall be nameless) where I had to explain the difference between hits and page impressions to the salesperson.    They do not publish figures because they are (presumably) nothing to write home  about. 

That doesn't mean that you should not advertise with them.  If you are selling software at £2million a pop then you are in a niche that can be very small.  Risk magazine is very prestigous and  the web site is very valuable real estate.  Alexa rating or not.

There are of course sites in our niche that have respectable figures.  Alexa is a way of sorting out the serious competition from the ones we can cast doubt on.  And because Alexa is free we can let the prospect see for themselves.

Bob Browning
Textor