They may be a bit cagey about their popularity relative to the almighty Google but there’s no disputing that Alexa does a damn good job of analysing domain popularity. This is quantitatively different from analysis of link relevance performed by Google. Amazon’s search engine is often useful as a mechanism of finding sources of knowledge, opinion or just plain stuff that are unknown to you but have been tried and trusted by others. Google is better at finding the needle in the haystack. The breadth of it’s search is great at detecting that one useful link from an ocean of fractured factums.
I guess the feature that excites (no pun intended) me most about Alexa is their Web Search Platform. Released in December of 2005, this enables developers to roll their own search engines, providing them with programmatic access to over 5 billion indexed pages through a nifty web service API. This is more than just a web API to retrieve basic alexa searches in a standardised form. It’s enables the creation of search applications which operate directly on defined subsets of the Alexa search space, consume alexa resources and flexibly publish results. DB programmers can think of it as the coolest stored procedure mechanism in the world today.