freejung - 5:25 pm on Sep 1, 2010 (gmt 0)

dvduval, I think you're on to something. I've experienced things that would support your hypothesis. For example, I recently stopped using a keyword group that had lower stats than others. Traffic for that keyword group immediately dropped. Sure enough, traffic across other keyword groups lifted slightly to compensate, with no overall loss of traffic (leading to a nice revenue increase, btw, because the other groups are worth more). And it's not that one particular keyword or set of keywords benefitted from the loss of the other - there was no spike in any particular keyword or group, just an overall slight rise in traffic across a large number of keywords.

As another example, there are a lot of cases where there are several versions of a keyword that are basically synonyms. The competition for each version is exactly the same - the same pages are in the top ten. However, I consistently rank better for the lower-traffic versions, despite heavily optimizing for the higher-traffic versions.

It feels like I have a fixed traffic allotment, and Google is balancing rankings across a large number of keywords to maintain it despite any changes I make.

There have been a lot of threads suggesting that Google is throttling traffic. There have been a lot of threads about on-site changes and additional content having little effect on overall traffic and rankings. I can certainly attest that my total traffic and from Google is remarkably stable, almost unnaturally so, and much more so than traffic from Bing/Yahoo.

The answer, then, is obviously to focus on building more domain authority, and not worry so much about particular keywords and rankings.