RedCardinal - 7:06 am on Oct 11, 2012 (gmt 0)
For any interested Cemper put out a report analysing 60 sites that were affected by EMD, 30 negatively and 30 positively. The report itself doesn't contain much useful analysis, but they do provide the 60 domains so you can dig further in order to try to classify diffs between EMDs that were hit versus those that actually improved visibility.
Taking about 20 of the sites here's some items I noted (hit sites | promoted sites):
Exact match keywords used in regular text (not internal anchors) on homepage (e.g. "Anchor Text"):
Multiple times (i.e. 6-10) | few times (1-2)
Domain name used in regular text rather than keyword ("AnchorText.com" vs. "Anchor Text"):
Little or no usage (<2) | Much more prevalent usage (>2)
The above could be signs of on-site over-optimisation, and a reliance on exact match keyword over brand.
Now the interesting part about links. While Cemper did come to the conclusion that links were the major element in EMD update, they really didn't back this up very well with their data IMO. Here's what I found via Majestic SEO (again Hit | Promoted):
Level of EM Keyword anchors (e.g. "Anchor Text"):
Generally all top anchors were EM | Top anchors much more diverse (similar to Penguin)
Linking root domains:
Very high number of EM keyword links from each linking domain | Low linking domain:anchor ratio (many hit sites seemed to be getting multiple EM links from each linking site)
Mainly EM keyword or commercial | Diverse anchors with far fewer EM
There's no way to know what mix of signals Google uses (other than to be certain it's complex), but my gut tells me it's a blend of anchor/link type together with some obvious on-site over-optimisation signals. Basically EMDs without SEO might still enjoy a ranking boost (as per the promoted set of domains in Cemper report).
If correct then diversifying inbound anchor text and reducing on-site optimisation might see EMDs increase rankings as Google refreshes the data.
Of course the above is all humble opinion, and as with everything SEO YMMV. Using the data set might be useful, and I think some users were looking for lists of sites they could analyse.
PS: one hit site (according to Cemper) that might be interesting to watch was the official Tourism Dubai website.