Forum Moderators: Robert Charlton & goodroi
It's number 3 but another page.
I know for a fact that the most compelling reason this (and another couple of pages) were no1 was inbound link anchor text.
I know it could be a 100 other reasons but just wondering if anyone was discussing this on the forum?
Has Google turned the dial down on anchor text?
I'm seeing it too, but haven't though of any other ideas for a fix than what you've mentioned. Although there was something strange that I noticed.
On a site I'm in charge with, the trophy phrase took a nosedive and it's now at 500+ on regional Google SERPs. Only one phrase. And only on the regional. And we did nothing at all recently, all links that could have been in the gray zone have been accounted for at least a month ago. So there's some changes going on.
Definately an anchor text problem, but I'd bet a reasonable amount of... hours of me doing research, that this is either the overuse of the same phrase, or being linked from off-topic pages or both.
If Google turned the dial up, it was for the filters, not the ranking mechanism.
...
On the other hand, the theory that they've added something to the ranking algo may fit as well... IF we'll start seeing changes for the other targeted phrases. For only a single one is affected, the one I'm assuming to be overused already.
Also, this particular keyphrase was hit only on a regional Google...
SERPs on Google.com were unaffected.
Meaning whatever this is, it's a LocalRank issue, at least for us. Our site has enough TrustRank while its LocalRank parameters were pretty low to start with ( not hosted in the country, not having the proper .tld, default is English which is not the language for the region... etc. ).
What do you see?
Is this happening asymmetrically?
Or are your rankings flying low on all / only the regional / only .com SERPs?
LocalRank is a specific patented method that involves taking the original set of results and reranking their relative positions to each other. The final ranking order is then based only on the links that are involved between those top 1,000 urls - and this method drops all other links. So it's "local" in the sense of "local within only the top ranking urls" not "local within a geographic area."
There are other algo approaches that take the top 1,000 "first stage" results and rerank them using other criteria than interlinking - many other approaches in fact. The recent slew of phrase-based patents list several of these - including phrase-based spam detection [webmasterworld.com].
When I see a single phrase drop dramatically, I always suspect some kind of final re-ranking step has been applied. But knowing the technical approach that's in use, and knowing exactly what factor is being measured that trips the re-ranking drop are two different things.
Also, this particular keyphrase was hit only on a regional Google...
SERPs on Google.com were unaffected.Meaning whatever this is, it's a LocalRank issue, at least for us. Our site has enough TrustRank while its LocalRank parameters were pretty low to start with ( not hosted in the country, not having the proper .tld, default is English which is not the language for the region... etc. ).
Here's a discussion of LocalRank...
Google's 2 rankings & you
New patent means new way of ranking
[webmasterworld.com...]
...and yes, as tedster points out, that word "local" can be confusing. ;)
Apart from the terminology of LocalRank, Miamacs analysis of his geo-regional drop is a good one.
I'm also seeing the filter dials being tuned constantly, and they can relate to link quality (including filtering designed to identify purchased links), anchor text patterns, apparent link-network patterns, phrase-specific duplication issues, etc etc.
Has Google turned the dial down in anchor text keywords?
Don't think so. My SERPs on various phrases keep going up fairly quickly, and the only changes (I'm aware of) are more keyword rich/phrase-matching anchor text links within my site to 'Related Content.'
However, they could be tuning the inbound/outbound links values (which I haven't been tracking).
p/g
If you pass these tests, then the anchor text weighting is fully applied, if you don't then you don't get the benefit.
Has Google turned the dial down on anchor text?
I don't think it's a mater of turning a dial up and down. More it's changing the weighting of certain words or phrases or combinations thereof.
Anchor text is considered in the phrase based patents.
Spread the inbounds around your site and vary the anchor text. The importance of anchor text is still as it ever was but its now easier to trigger a filter. Filters are also backward compatible, meaning what spammy linking schemes you did in the past will come back and haunt you. With this in mind you may even want to dilute the concentration of your allinanchor text and try and make the site a little more organic.
That's interesting, looking at the results for "widget design" on G.co.uk brings up a number of UK sites that are nowhere in the allinanchor for that term but do feature prominently for allinanchor "widgeting design"/'widget design uk".
Has anyone specifically covered this part of the phrase based patents?
That is just silly. Anchor tells SE's what a page is about.
Regional results for one, compare the allinanchor results for a generic term on the various English language Gs.
Here it goes again.
The site had LocalRank issues as in some of the startup links were from our own sites, most of which are hosted on the same IP. Not as the site itself, but relative to each other, hence I mentioned "gray zone". Most of these links have been accounted for already, and we didn't change a thing since then.
Also... and this is where the confusion lies... the site recently got linked to from a network of regionally optimized niche pages, which are heavily crosslinked and seem to form an entire network. AND adding to the problems, our site is not hosted in the same region, doesn't have the regional .tld, and is in English by default.
The fact that the pages took a dive only for the phrase used in these links, and only on the regional Google SERPs got me thinking, that perhaps LocalRank has been somewhat revised, and it doesn't only check IP ranges, but also networks, or "neighbourhoods". The mentioned network, our site, and the sites we linked to it from are all listed top 300 in the primary index. And if Google actually thinks that we're a part of this same-ownership network, regardless of a different IP being used, and pushed us down because of that, then we're in trouble. It may be science fiction, I'm not sure.
But I think I got myself cornered, there are just too many factors to consider. I'll need to eliminate them one by one.
And the reason why I don't think the dial for anchor text has been touched is that the site(s) I oversee rank all the same on .com for this phrase, and kept their position for everything else as well. If there's a change, it's in the filtering of same-anchor links, especially within a network that uses heavy crosslinking. ( Linking pattern analysis being extended. ) But I need more time to make sure.
The anchor text link that was affected was (KEYWORDPHRASE "REGION") - I had 5 links on one page on one of my sites (different keywords - different regions) pointing to 5 different pages on another site I own and closely linked in terms of what the sites are about - ie same sector - 95% 100% differing origional content. As I know everything about both sites (links to and from etc) it seemed to me change in anchor text filtering was a possiblity - although I know it is impossible to tell acurately).
These pages no longer rank for these terms although now some other of my site pages rank for the same keyterms but a place or two down in the serps.
Of course perhaps my new pages and all the other pages at the top of the serps just beat these old pages and this is the new order of things - who knows!
i agree with others, they wouldn't turn down the effectiveness of anchor's, otherwise it would completely change everything.
maybe one day the sites with the best content will rank at the top. they havn't quite gotten there yet. but are getting closer.
They could widen the factors included in the calculations though and the effects would not be too dissimilar, in the past they experimented by including file names, lasted about three months.
taking the game out of the engine and increasing the quality of the results. seems to be the end goal with a system that is so easily gamed currently.
You really have to wade through them and try to get a sense of what might be going on.
I'm not savvy enough to be able to tell you what is there but I do know that anchor text is considered probably in many ways. It's not just a matter of too many inbound and/or internal links having the same text. One phrase or word may get by with more identical anchor text than another phrase or word. And it may well be that one webpage or site may get away with more because they have stronger inbound links or some other factor that overcomes the filter.
You really have to think out of the box on this one.
