adder - 10:10 am on Sep 7, 2012 (gmt 0)
In my niche is a website that didnīt show up before penguin/panda and it has 80% keyword density. 100 of internal keyword link. Exact domain name and is a .biz domain. It is holding numerus #1-#10 spots on many different search terms. The ranked site are very similar but do rank.
Ok, let's elaborate on Martin's example by adding a real-life case study with a similar problem.
* An EMD site widgetry.tld wants to rank for a single-word keyword "widgetry". It is a medium-difficulty term in its home country which is a non-English market in a big European country.
* The site is obviously over-optimised. Homepage keyword density for "widgetry" is 6.65% and the other keywords on the density list is simple "noise" - words like; you, me, and, for etc... (whereas normally you'd also have variations like gadgetry, gizmos, thingies etc).
* Internal linking structure: the navigation has several dozens of links to sub-categories worded like: Widgetry Option 1, Widgetry Option 2 etc.
* Url structure: widgetry.tld/widgetry-option-subcategory/widgetry-product-type
So the url string will always have the keyword repeated 3 times!
* I've already found 100 low-to-medium-weight phrases that the site ranks for in its native Google.tld. It hasn't achieved any meaningful #1 positions yet but by ranking #5 - #10 for a string of good "sales" terms, it is by all means making some money.
* The spike in traffic occured some time between October and November 2011 although the site didn't add anything new during that time. I've looked at screenshots from 2010 and the site hasn't changed the content.
* Links-wise, the site has achieved links from 31 sites :) If we look at the anchor texts, 12 are branded, 12 are exact, 3 - longtail and 4 are variations. The link sources are mainly directories, bookmarks and only a couple of medium-authority sites.
* The results page for "widgetry" looks like this:
#1-#4 - massive international sites with millions of pages
#5 our spam example
#6-#9 variously-sized competitors of #5
#10 - niche articles compendum with a massive link portfolio
So the question is why and how? After seeing so many Penguin victims, I simply don't understand how was this site not taken down by Penguin? My only possible guess would be that it's the fact that 30% of it's anchor text is kind-of branded.