Forum Moderators: Robert Charlton & goodroi
When i read an article by the VLDB - Very Large Database Organisation:
http://www.vldb.org/conf/2004/RS15P3.PDFOur experimental results show that we can effectively identify a significant number of strongly reputable (non-spam) pages. In a search engine, TrustRank can be used either separately to filter the index, or in combination with PageRank and other metrics to rank search results.
Simply put, to establish "Trust" one must see [ or calculate ] a pattern of predictability. So, if you break that "Trust" you are likely to have to wait a long time to restore confidence , or , you are going to have to pack in a lot of extra / extra / extra good things to win back "Trust".
I kinda see this in the analogy of a Pyramid.
Why? - because a pyramid to me is built for the purposes of very precise outcomes, yet essential to it's engineering is the establishement of solid foundations. If you remove a brick from the middle, it will not alter it's course, nor the footprint of it's foundation. But mess up the footings in it's foundations and it's a hard job to change. But this is me talking concept - let's get our hands dirty and examine the detail from various angles.
So how do you repair "Trust"? How , when and will Google forgive the errors that you previously engaged in?
And what are the most biggest and most important elements of the "Trust" foundation that you must not mess with that will trigger the most unforgiving filters?
So how do you repair "Trust"? How , when and will Google forgive the errors that you previously engaged in?And what are the most biggest and most important elements of the "Trust" foundation that you must not mess with that will trigger the most unforgiving filters?
Nice thread idea. :)
I had a website that was hammered by Google.com in late 2004. Previously to this the website was top ten for many terms.
I would suggest this is what did me in:
Too much duplicate content
Issues with canonicalization
Too many outbound links
Unnatural link profile at the time.
Repetition of left side anchor nav
I can then tell you how I got the website back (it is now ranking like before, and has been since March 2007)
Reduced the number of outbound links was (and still am) VERY scrutinizing and picky about linking out. In general the source I link to must be high quality and useful for my visitors
Unified urls so that there was not source of the same content on my own website at different urls. Fixed any canonical issues.
Built one way links to the website over time and got links for a variety of sources including a few paid links from higher quality sources.
Asked a non-webbie friend to help me develop a better nav :)
Validated the site and checked and fixed any broken links.
Began writing new unique content on the website.
I think that the most important part of Trust in my case was who I was linking to. I was simply linking out to too many lower quality websites. To me this would send a signal of non-trust to G
I would like to think that most of those items above help establish trust and are likely used to some degree in rankings.
Two concurrent parameters were PageRank and ... whatsitcalled... Negative PageRank (?)
Which means the core analyzed sites either had a lot of links coming in from OK sources,... or had a record high on-topic concetration of links going out (!) to such, without too many people linking in ( outbounds in combination with content, and actually used by people so directories didn't play a role ). Some people have reported that linking to authoritative resources had stabilized their rankings, and even if a very subtle way, I too can conclude that for whatever reason ( might not be algorithmical at all, just an indication of user behavior ) Google has favored sites that if linking out, linked out to safe, well known, on topic sites. Not to competing pages, but to additional information on the greater- or subtopics.
With all that said, pages that don't have outbound links, especially not with their own keywords, always rank better.
...
It was usually normalizing anchor text in the navigation - to be more user friendly - that helped on the sites that I repaired. Obvious errors are a no-no that's for sure, but then again they're tolerated on highly trusted domains. Many things, like broken HTML and dupe content has been slipping through, yeah, well some pages dropped out, but they haven't been really harming an otherwise legit and popular site. But if the anchor text was targeting the wrong things, or the right things but in an unnatural manner ( stuffed keywords, wrong word order, lacking relevancy ... etc. ) it did hold back the effects of being trusted.
So in my experience, if a site would otherwise be high on TrustRank, yet it's not performing well - almost as if it was penalized - it usually has something to do with relevancy, near-spammy, repetitive, broken TITLEs, 'unfriendly' anchor text in the navigation, artificially keyword-injected content or accessibility problems. Basic HTML errors and such have never been an issue from where I'm looking at it. Not to the extent of sending down an entire domain a few hundred positions at least.
( ...except when their effects caused any of the above mentioned problems )
...
If you compare two pyramids, say, of equal height, a wide-based pyramid is inherently more stable than a narrow-based pyramid. And, for a given base size, the higher a pyramid becomes, the less stable it is.
A pyramid whose base is narrow compared to its height can tip over with only a very slight lateral motion at the top, something which might not require a great deal of force... particularly once the top of the pyramid starts moving. A very wide-based pyramid is unlikely to tip over at all.
In terms of the metaphor, I leave it to you to fill in what the pyramid's height and its base size might represent.
Sure, the foundations include basic accessibility, a crawlable site, no repetitive, spamlike practices, no malware installing itself upon visit, no overused tags, and so on, but in general, the relevancy your trust supports, or rather, the trust your relevancy supports is the trusted-theme pyramid. You can't really separate the two anymore.
Some months ago I've posted the findings of my research on the -950 penalty for example ( long forgotten on that thread though ). Acquired and used half a dozen domains, and kept on helping others here too, which all added to the data. One of the conclusions was that a site may have - by this example - a theme pyramid that is out of balance.
The navigation that isn't spammy, but isn't irrelevant either ( that 'underoptimization' trend freaked me out ) is the framework you'll need. It's more like a wooden/ steel structre but, whatever, that is the indication of the direction you'll build in.
Sites with -950 for given phrases had a structure out of balance, with its height being sufficient in some areas ( ie. relevancy combined with trust reaches a certain threshold for one set of keywords ) but relevancy lacking at others. The height in my opinion is the popular/competitive nature of any given topic. Which you need to support with additional relevant content, inbound links' anchor text, pages to have substance. If the building climbs high enough, it reaches the required-trust threshold to appear in 3, 2, 1 word competitive searches.
Some sites have narrow, one-block-wide structures, some have multiple pyramids built together. When you wander off-topic you're trying try to place a block in mid air. The same height as the highest point of the structure... yet it's not supported by what you've built so far ( ie. your site is relevant and trusted for a top keyphrase, and you aim for something equally competitive, but 'off-topic' ).
That piece will fall right where it belongs to... the level of the lacking foundations of a new theme ( -950, buh bye... )
If the site was not trusted, it would drop out of the SERPs entirely.
Some sites have a wide enough top-level on their pyramid to support new themes, even phrases they never used before. For example if the the domain is known to be a well trusted source for a top-level category ( ie. 'news' in general, and not just 'widget news' ) it can, at will, post new stuff about nearly anything. Might end up in a different 'vertical' though... ( for the time being ).
Basic level trust is aquired easily, regained easily all the same. Technical related errors and hijacked pages on sites I've seen - with an otherwise pure white history - had always regained it fast, no matter what happened. Unless they were like indexed and left in the index with completely unrelated / offensive material for a long time. But that again, was relevancy based filtering, looking at the content, and not just 'trust'.
The mid-structure, the most important of them all, relevancy was always harder to build, and to build both fast and SAFE enough to hold the next level of keywords. And with every round of Google tinkering / updating their understanding of natural language, the content and inbound link text you needed for it had to be more and more... uh... natural. Ok, at least not 'unnatural'. ( 'mycity my city widget widget widgets' )
Also, you can't leap from peak to peak unless you have the foundations to do so. And you shouldn't build the structure too high and narrow for it'll fall ( same anchor text can get you penalized for months... I've had a site where this happened way too soon, and by the time I could get it to work again I had so many links with so many variations that all the OTHER phrases were bringing in a lot of traffic already )
That's my insane vision of a 'trusted-theme' pyramid. I really can't say I've seen any other part as hard to 'regain' the graces of Google in, than the trusted-theme, relevancy related penalties. Those are the times when your trust might show a virtual 'soon to be built' ( or repaired ) image of a structure... by including the given phrase at -950. Or not, for it could fall out entirely. If you support it, it'll suddenly appear in top positions. The bricks are relevancy, both on and off-page, but quite obviously ( we're talking Google here ) the best among them are the anchor text of high quality links.
...
[edited by: Miamacs at 1:26 pm (utc) on Nov. 20, 2007]
