indyank - 6:57 am on Aug 31, 2011 (gmt 0)

My understanding is this - Google uses keywords as the primary unit (starting point) for applying panda. A certain percentage of the top traffic driving keywords are picked and sites with pages that rank high for those keywords are evaluated.

I also think that they might or might not be evaluating the page that pulls in the traffic. But they certainly pull a few other pages from the site (domain), either randomly or based on some relativeness criteria, and evaluate signals emanating from those pages.A quality score is computed and a site wide drag is applied to all pages on the site. The strength of this drag is determined by the quality score.

When you break a site into several smaller sub domains, they would probably evaluate a small sample of pages from the sub domain that hosts the traffic driving page. If the site happens to have good content for pages within that sub domain, pages within that sub domain have higher chances to recover. Even if you have some low quality pages on that sub domain, a quality score is determined for that sub domain and the damage is limited to that sub domain.

At the same time, there might be other sub domains which wouldn't be evaluated at all as they don't have pages for the traffic driving keywords that panda uses as a trigger. Since pages have moved to their new sub domains, they will have newer URLs and panda might not evaluate such sub domains and their pages for quality. However, they start getting the pre-panda traffic, as all other page strengths are transferred through the 301 redirect.

[edited by: indyank at 7:33 am (utc) on Aug 31, 2011]