Thread: G Page Rank woes

Originally Posted by link8r

Our theory - it's got to have changed. PR must be relative - and in context of - the search index its dropped into. For example, take a word that is not used on the web a lot, like a name. Register that as a .com - e.g. DavidTJohnson.com. Build 20 backlinks to it, put up some content, pages, - do all the seo stuff right. chances are within a year it will be PR4. Change the page title to "Property Ireland" - I bet you $10 it'll drop to PR1.

I can't see how a website can have 30 backlinks in a narrow category and a PR of 4 can also hold the same PR in property, holdiays, car rental or other high search categories. Note: I'm talking PR, not SERP.

so a couple of thoughts on this - pointing 30 links at siteA (niche A) and the same links at siteB (niche B) is obviously going to have different effects. It's very likely that Google passes more weight across links from other sites in your niche, and this may also affect the flow of pagerank across those links.

Comparing 30 links pointed at siteA and 30 links pointed at siteB is like comparing apples and oranges. Even if the 30 links pointing at each target come from the same 30 sites it's possible some pass more weight to one of the targets over the others.

The other thing I'd ask you - are you talking about editorial links or links in footers, sidebars etc? I think I remember you mentioning about turning on hundreds of links from your own sites previously. Again 2 links in a footer from siteA to siteB and siteC. Sites A and B in same niche. Site C not. The link to site C may tip the threshold for spammy link while the link to siteB might not. That would affect PR flow also. Not suggesting you have spammy sites or anything, but just making the point that testing with site networks has inherent flaws.

Originally Posted by link8r

Perhaps your site got assessed and relative to it's neighbours within it's category, it's got less backlinks/points for it's PR ?

I have too many sites that have 3 or 4 backlinks as sumtotal with huge PR, some even as high as 5 or 6. I've taken on many as SEO projects and watched the PR drop once you change the page title from "ACME CORP LTD" to "Acme Cars Dublin" without chaging BL or even increasing it.

How have you watched this? Up until quite recently Google only exported PR once every 3 or 4 months. How can you be sure that changing title affected PR? How can you be sure that Google didn't just stop counting links that you pointed at those sites? All I can say is that I've never seen nor heard of anything like this, and I'm yet to hear your response to my previous question on your theory - if this is true then does the converse hold? If I change from "Acme Cars Dublin" to "ACME CORP LTD" will my PR increase?

Rgds
Richard

Just want to add in the quote:

If content didn't make a difference then both domains should have an equal PageRank after an equal amount of time with equal backlinks. The backlinks I create, I have a couple of hundred I turn on, so they are almost always the same, almost always from similar PR pages. My observation is that the domain without a highly competitive phrase doesn't need the same backlink count to get a higher PR.

That's from the other PR thread. The ability to turn on/off backlinks makes them non-editorial IMO. Very likely some of those links pass value dependent on where they are pointing. Pointing 200 links at siteA and siteB doesn't mean that the same value should also flow. I'd also be very careful basing your theories on what happens within your own network - it has it's own dynamics, and may not e representative of global reality.

Rgds
Richard

Originally Posted by RedCardinal

so a couple of thoughts on this - pointing 30 links at siteA (niche A) and the same links at siteB (niche B) is obviously going to have different effects. It's very likely that Google passes more weight across links from other sites in your niche, and this may also affect the flow of pagerank across those links.

Why would more weight be passed to SiteA than SiteB for example? If the links are in a similar position on the pages? The relevance of oldsite1 and oldsite2 must be passed equally to both?

Originally Posted by RedCardinal

The other thing I'd ask you - are you talking about editorial links or links in footers, sidebars etc? I think I remember you mentioning about turning on hundreds of links from your own sites previously. Again 2 links in a footer from siteA to siteB and siteC. Sites A and B in same niche. Site C not. The link to site C may tip the threshold for spammy link while the link to siteB might not. That would affect PR flow also. Not suggesting you have spammy sites or anything, but just making the point that testing with site networks has inherent flaws.

Yip, it's not exactly a controlled environment. And having only access to the PR on a 1-10 scale is a bit blunt.

I'm going to back to the criteria for the observation is on sites with low inbound links. Where a site has a PR of 4 but has less than 10 inbound links. We're talking fairly rubbish links here - probably no anchor text, no great relevance. The typical ones would be from the website designer (portfolio), the domain registration watch sites, the odd directory link, a link from a site owned by the owner.

I guess you'll be able to take my point from this and give your thoughts on this: If the PR is exclusively down to backlinks and the content on the pages those links are coming from, then how do these sites develop a PageRank of 4/10 ? Within a year.

Originally Posted by RedCardinal

All I can say is that I've never seen nor heard of anything like this

Nope, there's not a lot of information except on 3rd party websites, like Wikipedia, but then all such sites cannot be produced as gospel, so it's pointless quoting them. There is very little information from Google itself. It's an observation we've noted over the past 3 years or so. We watch it because simply put we record the PageRank when we report back to the clients - and most of the time, they're curious about the PR drop. This coul be 1 month or 2 months or 3 months afterward - depends obviously when the PR export is done. We check the PR on the domian name and the index file (sometimes www.mydomain.com and www.mydomain.com/index.html) have different PR's in the toolbar. Toolbar guessing?

Originally Posted by RedCardinal

and I'm yet to hear your response to my previous question on your theory - if this is true then does the converse hold? If I change from "Acme Cars Dublin" to "ACME CORP LTD" will my PR increase?

I missed this, apologies. I've never gone back, so I have no observations for it ... its a good question. I don't know. I've recently discovered just how strong the Google memory is for keywords it associates with sites, its very good. Even if you change the site. So I guess it might depend on what you move back to.

Originally Posted by RedCardinal

That's from the other PR thread. The ability to turn on/off backlinks makes them non-editorial IMO. Very likely some of those links pass value dependent on where they are pointing. Pointing 200 links at siteA and siteB doesn't mean that the same value should also flow. I'd also be very careful basing your theories on what happens within your own network - it has it's own dynamics, and may not e representative of global reality.

True. I'm only dealing with 100's of sites, not tens of thousands, it's a very small sample rate. I've mentioned it to other SEO's that I know and they agree. Doesn't mean it's right!

I've read this a few times - PageRank reflecs our view of the importance of PageRank by considering 2 billion terms - so there is a relationship between page rank and the terms. It doesn't state if the terms are taken from your page or the page linking to you only.

PageRank Technology: PageRank reflects our view of the importance of web pages by considering more than 500 million variables and 2 billion terms. Pages that we believe are important pages receive a higher PageRank and are more likely to appear at the top of the search results.

My interpretation of this is the that PageRank is more than just backlinks - it's also about which search terms your site is relevent to. So PageRank is how important your site is in relation to other sites for a category/term/phrase. Thus being in a bigger directory means you have to have more of the other PR factors in effect in order to have a higher PR.

Originally Posted by link8r

Why would more weight be passed to SiteA than SiteB for example? If the links are in a similar position on the pages? The relevance of oldsite1 and oldsite2 must be passed equally to both?

Absolutely not. Google discounts many links. Probably in fact the majority of links it finds. Link#1 from SiteC to SiteA might be relevant and included, while Link#2 from SiteC to SiteB may be less relevant, trip a signal, and be ignored. Link#1 would carry weight and PR, while Link#2 may not.

I'm going to back to the criteria for the observation is on sites with low inbound links. Where a site has a PR of 4 but has less than 10 inbound links. We're talking fairly rubbish links here - probably no anchor text, no great relevance. The typical ones would be from the website designer (portfolio), the domain registration watch sites, the odd directory link, a link from a site owned by the owner.

I guess you'll be able to take my point from this and give your thoughts on this: If the PR is exclusively down to backlinks and the content on the pages those links are coming from, then how do these sites develop a PageRank of 4/10 ? Within a year.

Just my opinion, but I think that what Google has been showing in the toolbar has for quite some time been bastardised. I think it will only be a matter of time before it is complete rubbish - that would be the easiest way for them to retire TBPR. Just my imagination at work, but I'm pretty sure they'd love to retire it, but the problem is they created this monster by basing their marketing around PageRank.

We watch it because simply put we record the PageRank when we report back to the clients - and most of the time, they're curious about the PR drop. This coul be 1 month or 2 months or 3 months afterward - depends obviously when the PR export is done. We check the PR on the domian name and the index file (sometimes www.mydomain.com and www.mydomain.com/index.html) have different PR's in the toolbar. Toolbar guessing?

Technically that means you have canonical problems, but I'm pretty confident that Google sees www.example.com/index.EXT and Example Web Page as the same page where they should. This change was introduced about 1yr ago. That's not to say there may not have been other issues in some of the cases you refer to however.

I missed this, apologies. I've never gone back, so I have no observations for it ... its a good question. I don't know. I've recently discovered just how strong the Google memory is for keywords it associates with sites, its very good. Even if you change the site. So I guess it might depend on what you move back to.

So the biggest problem with some of your theories as see them, and I'm not trying to say I'm right and you're wrong BTW, is that it is very difficult to asses causality in such massively dynamic systems. With the amount of data points Google is using for it's calculations, what you and I think we see, and ascribe causality to, is more likely than not even a factor in the outcomes observed.

Originally Posted by link8r

I've read this a few times - PageRank reflecs our view of the importance of PageRank by considering 2 billion terms - so there is a relationship between page rank and the terms. It doesn't state if the terms are taken from your page or the page linking to you only.

What makes you sure that 'terms' == keywords. Google spreads lots and lots of FUD in order to mask what it does.

My interpretation of this is the that PageRank is more than just backlinks - it's also about which search terms your site is relevent to. So PageRank is how important your site is in relation to other sites for a category/term/phrase. Thus being in a bigger directory means you have to have more of the other PR factors in effect in order to have a higher PR.

Personally I still believe that PR is purely driven by the PR of the links that point at a given resource. The changes I think they have likely rolled in revolve around which links they count, and how they weight those links. That's purely based on what I hear Googler's saying. No tests, and likely no way to test that thesis.

Originally Posted by RedCardinal

Absolutely not. Google discounts many links. Probably in fact the majority of links it finds. Link#1 from SiteC to SiteA might be relevant and included, while Link#2 from SiteC to SiteB may be less relevant, trip a signal, and be ignored. Link#1 would carry weight and PR, while Link#2 may not.

Can you go into more depth on why would link#2 from SiteC to SiteB be less relevant than the link#1 to SiteA

Originally Posted by RedCardinal

Just my opinion, but I think that what Google has been showing in the toolbar has for quite some time been bastardised. I think it will only be a matter of time before it is complete rubbish - that would be the easiest way for them to retire TBPR. Just my imagination at work, but I'm pretty sure they'd love to retire it, but the problem is they created this monster by basing their marketing around PageRank.

Probably, given that so many SEO's don't believe in it's relevance.

Originally Posted by RedCardinal

Technically that means you have canonical problems, but I'm pretty confident that Google sees www.example.com/index.EXT and Example Web Page as the same page where they should. This change was introduced about 1yr ago. That's not to say there may not have been other issues in some of the cases you refer to however.

And you'd be right. I've no idea why I posted that now.

Originally Posted by RedCardinal

So the biggest problem with some of your theories as see them, and I'm not trying to say I'm right and you're wrong BTW, is that it is very difficult to asses causality in such massively dynamic systems. With the amount of data points Google is using for it's calculations, what you and I think we see, and ascribe causality to, is more likely than not even a factor in the outcomes observed.

If you mean the theory that changing a page title and/or meta tags can cause a PageRank to drop - probably would be very difficult to prove.

But can you explain how a website with 1-3 backlinks can have a PageRank of 4 or 5 if BL is the only factor?

Originally Posted by link8r

Can you go into more depth on why would link#2 from SiteC to SiteB be less relevant than the link#1 to SiteA

SiteA = entertainment site
SiteB = car hire site
SiteC = music site

Link#1 (C->A) relevant
Link#2 (C->B) not relevant

Google likely filters out link value by the relevance between linking and target sites.

If you mean the theory that changing a page title and/or meta tags can cause a PageRank to drop - probably would be very difficult to prove.

But can you explain how a website with 1-3 backlinks can have a PageRank of 4 or 5 if BL is the only factor?

Set up a domain and get a link from Adobe HP and no other links. I'd be surprised
if that domain doesn't get TBPR6 or TBPR7 next update. But the main reason is that I'm quite sure that they've been exporting FUD in the toolbar for quite some time.

I'm pretty sure I've mentioned this opinion a few times in the past.

I asked how a site could have a PageRank of 4 or 5 with rubbish links - not from Adobe or HP .... that was the basis and context for my argument.

Having a PR6/7 from a link from Adobe etc makes sense.