OK. So the other day I published the first major new article in a while. It was mostly a combination of various older articles. It was a big keyword hit, very popular keyword phrase in the title, headers and all over the shop. Aim: to get to page 1 for this keyword.

So, the page went live late on the 13th Feb. The following day it received and expected amount of traffic based on the traffic to the old articles. On the 15th it received a big boost from Google which has since declined. Of the 408 keyword searches for this page, 362 were (not provided), 89% of visits. It is back to a normalish 8.8% today.

So, my wacky warehouse theory: Google is using (not provided) to test new pages. It will open the page up to a bunch of keyword search phrases for a day, maybe in a small test group, to see how users treat the page. If bounce rate (that loads of people keep talking about) is not above the threshold then it gives it the green light to then makes its way to the top of search after some lucky SEO work. If people do not like it, it is relegated to the depths of the index, labelled as "probably just keyword spam",

I had 3 hours sleep last night. That is my excuse if this is completely daft.

Edit: Oh, the other wacky theory that this is related to is Panda has been Caffeinated. A few times I made changes and saw almost instant results (improvements) to rankings.