Forum Moderators: Robert Charlton & goodroi
-950 penalty [webmasterworld.com] - Usually seen for a specific search term and URL
Think of this as an "end-of-results" phenomenon, rather than any specific number of the penalty. What we see is a URL that was previously top ranked for a certain search now showing up on one of the last pages of the SERPs. Most of the time this only affects specific searches -- on other searches that same URL can still be first page. The low ranking sometimes comes and goes with various new data pushes from Google. It's a very odd thing to be #1 and then #896 and then #1 again!
The reason that our discussion of the -950 is now over 2,000 posts is that we still don't have any definitive understanding for it. Some people have recovered after they "de-optimized". In particular, they lowered the occurances of keywords, especially in anchor text but also in other areas of the page. Other people have recovered after acquiring a good backlink that uses the problem search term. And yet others have added some semantically related text to their page and seen a recovery soon after.
None of these approaches has worked consistently, as far as we can tell, and so we cannot say that people recovered BECAUSE of those changes, only that they recovered AFTER those changes. And (this is most frustrating) many people who are suffering from end-of-results have not seen the situation change at all, no matter what they try.
(The most recent part of the -950 discussion is here [webmasterworld.com])
Minus Thirty Penalty [webmasterworld.com] - Affects the entire domain on every search
When this penalty situation was first noticed, it worked in a very odd way. Every search that would normally bring up a url from that domain now showed that url on the top of page 4, position #31 precisely -- and this was even true for a search on the domain name itself.
The penalty has mutated a bit since then -- the domain name search, for example, is apparently not always penalized today. Also, the depressed position is not consistently at #31, but every search is still pushed down by about 3 or more pages.
This penalty appears to be manually placed, not by algorithm. And in almost all cases where it has gone away (one recent report differs on this) the "minus thirty penalty" has been manually lifted. The minus thirty penalty seems to indicate a severe loss of trust by Google in that domain.
I don't want to have a side discussion about either of these penalties, so let's take further discussion into the respective threads, linked above.
(The most recent part of the Minus Thirty discussion is here [webmasterworld.com])
[edited by: tedster at 2:23 am (utc) on July 6, 2008]
First, Matt Cutts has commented, very briefly, on Search Engine Roundtable [seroundtable.com] about what we're seeing.
annej, regarding the -950 thing, I'd watch this video I made: [video.google.com...]
Starting around 1:42 into the video is where I talk about this.
The link that Matt gives us above goes directly to the 1:42 spot, and his relevant comment is pretty short. He moves quickly to talking about the Big Daddy infrastructure.
That's a good video to watch for other reasons, too - he talks about how and why the estimated number of results can be so far off, how many changes get pushed live every week, and how IP addresses are not always locked into any one data center.
-----
Experience has shown that there's no one action or set of specific actions that will always work to lift the -950.
I first brought up a connection between the -950 and the phrase based patents in Feb 2007 [webmasterworld.com]. Everything I've seen since then has confirmed for me that there is a mechanism in play that is based on the logic of the spam detection patent [webmasterworld.com].
Key points that can make this -950 difficult to "fix":
1. The re-ranking is triggered by crossing a threshold.2. The threshold can be different for different search terms.
3. The threshold can be different for different markets or website taxonomies.
4. The threshold is set by measuring and combining many different types of
mark-up and grammatical factors, and not by absolutely measuring any
one factor.5. The threshold is NOT set absolutely across all web documents. So phrases
in the travel space can be held to a different measure than, say, phrases
in jewelry e-commerce.
The patents suggest scoring all kinds of areas, for example:
"[0042] ...grammatical or format markers, for example by being in boldface, or underline, or as anchor text in a hyperlink, or in quotation marks."
"[0133] ...whether the occurrence is a title, bold, a heading, in a URL, in the body, in a sidebar, in a footer, in an advertisement, capitalized, or in some other type of HTML markup." Note that measurements are suggested here for position on the page.
Going over the top with a "de-optimization" effort could deflate your pages to the point where they NATURALLY should rank at 950! So use a gentle touch, record your changes - and know that if you are just barely over some threshold then it might not take much to move you back.
Also, the threshold will be re-calibrated from time to time and you might "pop out" of the penalty without doing anything at all, or for reasons that are not related to the changes you did make.