freejung - 6:36 pm on Sep 1, 2010 (gmt 0)
Do you mean you've stopped writing new content targeting these keywords (i.e. using them in titles of new pages) or you've removed old pages that were optimized for those KWs.
Neither. I mean that I've removed these keywords from the titles (and in some cases even the body content) of pages that used to use them.
I talked about this in another thread -- in my niche, there are words (widgets, woozles, wingdings, doodads, let's say) that basically mean the same thing, but with slightly different connotation. However, use of "doodads" in a search string generally signifies a slightly different user intent (what they intend to use the doodad for), and that intent is significantly less useful to me than the other possible uses for my widgets.
So before I might have a page titled "Red Widgets | Red Woozles | Red Doodads" (that's an oversimplification, but it works for illustration) and I replaced the title with "Red Widgets | Red Woozles | Red Wingdings"
Now, here's the part I find interesting. I would have expected this page to decrease a lot in traffic for terms involving doodads, and increase a lot for terms involving wingdings, and basically stay the same for the others, right? But what happened instead is that the page increased in traffic slightly for all of the other keyword groups - and this happened across many pages all at once. Actually, traffic for widgets (the most popular variant) increased slightly more than traffic for wingdings. The overall traffic gained pretty much exactly compensated for the traffic lost, hence the speculation that there is some sort of allotment.
However, in the course of explaining the details, another explanation just occurred to me. It could be that Google understands that widgets, woozles and wingdings are more closely related to each other, in terms of specific meaning and intent, than any of them are to doodads. So getting rid of doodads is making me more semanticlly relevant for all of the others at once.
That seems a more plausible explanation, however it still seems odd to me that the overall traffic balanced out so well, particularly because traffic numbers for the various synonyms are radically different (doodads gets a lot more traffic than wingdings or woozles, though not quite as much as widgets).