potentialgeek - 4:37 am on Jul 23, 2008 (gmt 0)

> How long have your sites been recovered for?

June 4 was the breakout day. Ironically at the same time when other sites dived. I think a month is reasonable amount of time to see changes. Google seems to do most of its stuff (updates) in a month or less.

> I might be stating the obvious / already stated here but I may have another clue. On one of my sites where some (20% max) pages are 950edI have noticed that most of these pages are not cached, and whats more whilst they have tons of unique contect, the subject is very similar to the home (eg a complete overview / explanation of 1000 words).

Similarity is one of the issues webmasters suspect contributes to the 950 penalty or the Phrase-Based Spam Penalty. One Google patent seems to indicate the idea behind it is spammers target like a gunshot v. rifleshot, aiming to saturate a wide swath of similar words and phrases (e.g. instead of just car, also vehicle, auto, automobile, etc.). So project that to a large topic and every related search phrase... lots of very similar phrases... that could get you 950d/penalized.

Incidentally, Matt Cutts recently made a comment that makes me wonder if Google is more sensitive to phrase repetition aka keyword density than it used to be.

When weaving keywords into a main page, Cutts says, some zealous Web publishers will use the term over and over again. That's called "keyword stuffing." It's a big Google no-no that can have your site removed from the index. "After you've said it two or three times, Google has a pretty good idea — 'OK, this page has something to do with this keyword,' " he says. "Just think about the two or three phrases you want to be known for and weave that in naturally."

(I recently got penalized for a page that used a phrase more than two or three times. I removed the repetition and bounced back completely.)

> Remove pages with little content on them and merge them where possible to other pages.

I absolutely agree. Thin pages are high risk pages especially when there are lots of links but little content. Spammers generate millions of pages quickly that are worthless. If you are creating thin pages which aren't really justified (why a separate page if it has almost nothing on it?), you're at risk. Same again, if you put a lot of links to a page with very little content. The Google algo has to wonder, How can that be justified? Why are they linking to nothing? Unless it's to generate more internal links to a page in the hopes of higher ranking?

> 6) I do have a related article section below articles. I haven't removed them as this is keeping us alive with traffic flow and visitor retention. This related article section of course uses the title of the article which is in the h1 tag of the article. Tried different variations, mixes, similar but not exact words/phrases.

This you might need to change. I think my changes on related articles freed me from the 950. I had dupe anchor text on the same page. I didn't mess around with changing the anchor text except to use an icon with a link, and just stripped all the anchor text of the link, so it still appears as text right beside the icon. With related documents you can just use a small file icon (like many news sites). It's so common it won't confuse users.

Related articles

[icon+link] Article title [no link]

p/g