The short questions are:
how do you justify SEO investments, and how do you track them?
This is not a question about how to track search traffic. We spent six figures on Omniture, and now have plenty of pretty graphs and trendlines.
No, the question here is whether there are any best practices regarding search-traffic tracking to justify to CxO's that an investment in SEO is worthwhile.
Here's the challenge -- we have a multi-million page website with lots of user generated content. If something on our site gets popular in the blogosphere, we get lots of inbound links and our search traffic goes up. If Google adjusts an algorithm, our search traffic goes down (heh, never up), regardless of any "SEO improvements" we may have made. These are not subtle swings.
Against that background, we have to decide whether to invest resources in improving onpage factors, nav issues, linkbuilding, content development, etc. The folks on Mahogany Row want to see a projection like "if we spend $X we'll get Y% increase in search traffic." Good enough; I can work up ridiculous projections just as if I went to Marketing school.
But how do we track that? There's no baseline curve against which improvements can be measured. Search traffic varies day to day.