Content_ed - 6:21 pm on Dec 6, 2011 (gmt 0)
I should have put more detail into the post, let me clear up a few points.
First, the traffic gain from Google for these pages was around 5X, quintupled.
The pages were moved from a site with around 150 pages total to a site with around 30 pages total.
I didn't mean Panda would literally recognize the content on the pages, I meant Panda might associate them with the old domain through the 301's. I don't know whether Google keeps a record of 301's or uses them and forgets them. And I have to leave them on the old site in any case for the incoming visitors not using search.
And while you may argue this is subjective, there are no "problem" pages on any of my sites. No syndication, no community content, no product reviews, other than the infringements, it's all unique, it's all professional.
But I do have a crow-eating theory I've been looking at for a few weeks that might explain Panda for my two Pandalized sites. I think everybody can agree by this time that there's no one-size-fits-all factor, they rolled out several major changes in Pandas.
I'd always believed that multiple subjects on a site is good or at least neutral. I always distrusted single subject sites myself as SEO garbage and expected search engines to do the same. WikiPedia, Yahoo Answers, Amazon, some of the hottest sites on the web are clearly multi-subject.
But here's a bit I never considered until talking it over with a friend recently. Very large multi-subject sites will tend to have an even distribution of pages on different subjects and of incoming links to those pages. It's not like all the links to Amazon focus on Harry Potter books or all the links to WikiPedia are for spinal curvature. It wouldn't surprise me if they obey the standard Pareto distribution, 80% of the incoming links and traffic going to 20% of the pages. And maybe, just maybe, Panda has tried applying a rule like this to determine website quality. It would certainly favor the big brands.
My worst Pandalized site, the one I just moved the pages off, is flipped very badly the other way. Without importing all of the pages into Word to count words, I would estimate that 5% of the pages on the site accounted for more than 95% of the incoming links and more than 95% of the traffic. WebMaster Tools keywords report includes none of the keywords from the popular content in the top ten for the site. And the subjects are entirely unrelated, about as close as Civil War Reenactment to is to nuclear power generation.
Before Panda, not only didn't it matter, it aided in some accidental discovery on the part of visitors who weren't looking for the other subjects but were engaged once they found them. After Panda? Maybe this is our greatest sin.