Backlinks - quality, not quantity

I, Brian · 10-31-2003, 02:42 AM

For a while I thought that signature links were an easy way to build up PR and - especially - get good anchor text to help page rankings.

However, after a few months experimentation, signature links are pretty worthless as backlinks.

Not only are the links usually poorly weighted, so that PR transfer is extremely limited - but, also, Google does not appeart to care for the number of backlinks from any site, as much as the number of different sites being linked from.

Not only that, Google seems to be clustering themes - ie, for your backlinks to be of any real worth, then your backlinks must be from other sites covering the same or similar topics.

Although signature links are a good way to form anchor text, without being applied in content then as site backlinks they will have almost no effect in terms of either their number or PR movement.

Signature links still remain an important way to direct human traffic to a sites. However, they are, in effect, almost worthless for Google rankings. Google insists that quality, not quantity, is more important.

PhilC · 10-31-2003, 03:06 AM

Sig links are on pages, often low PR pages, that contain many OBLs, so the PR gain can't usually be all that much, and it would take a great many of them to make any noticeable difference. Also, many forums hide the links from spiders, even when the Status bar shows the URL, and some forums aren't spidered at all.

It would make excellent sense for Google to drop sig links and similar, but it wouldn't be easy to spot them programmatically. It would be much easier to spot repetitive links, like those at the foot of every page in this forum, but not so easy to spot sig links. And yet John got #1 for "RV1 servers", simply by placing that link at the foot of every page in this forum and someplace else - a helluva lot of repetative links, but links of the type that Google could devalue or drop much easier than sig links if they wanted to.

I don't know what your experiments invloved but there is good evidence to show that repetitive links do count. Google may devalue or drop repetitive links - I've put that theory forward myself, but they still appear to count even if devalued, and if they count just a little bit, then they are worth having.

burgi82 · 10-31-2003, 04:00 AM

Hi,
I think that sig links can still be very useful. I have one site, which has a PR of 6, and only has like 4 links that have PR 4 or higher.
The rest is from sig. links at a few other forums.
So, sig links do work, but as mentioned before, you need a ton of them because forum pages usually have a low PR and there are tons of links in each thread generally.
Also, I agree that repetitive links still do count.

I, Brian · 10-31-2003, 04:42 AM

Quote:

Originally Posted by PhilC

I don't know what your experiments invloved but there is good evidence to show that repetitive links do count. Google may devalue or drop repetitive links - I've put that theory forward myself, but they still appear to count even if devalued, and if they count just a little bit, then they are worth having.

Certainly the issue of "theme/context" seems relevant. In this case, I would hardly be surprised if this forum ("internet", "marketing", "webmaster") was seen to be "on topic" for the subject of "servers".

HOWEVER, an important point to note is that John has ranked well for the term "ev1 servers" - which is a different ball park to "ev1.net servers".

At the end of the day, my argument is simply that almost randomly "link spamming" ie,

1/ posting in forums for the sake of posting only to get the sig link(s) up
2/ and also extensively cross-linking a network of domains -

are fairly pointless endeavours, and appear little rewarded.

Of course, I am mooting the point for discussion accroding to experience with my own sites - especially with interlinking my own domains. There was a PR shift, but little movement in rankings according to number of backlinks when off-topic.

In summary: A high number of backlinks from many different on-topic domains are far more valuable overall, than a high number of backlinks from a few off-topic domains.

PhilC · 10-31-2003, 04:53 AM

I agree with you that themes seem to matter, although I'm not totally convinced that themed groups of pages matter all that much. The only evidence I have for any theming is what B&P wrote way back about the text around a link having some relevancy to the link itself, which in turn is very relevant to the target page. So it's one of those things that, because it's there, use it. I create themed groups of pages, and I create text on the linking page to suite the target page - because it's there to be done, and I'm sure that some of it has some effect.

I don't think it's a case of this forum being on-theme for "ev1 servers", because it isn't. Only a very small part of the forum is about servers. Most of the stuff here is about other topics. But that's the phrase that is used as repetitive link text on tens of thousands of pages over 2 websites.

I, Brian · 10-31-2003, 04:56 AM

Sorry - I edited my post after you replied without realising.

As for themes - if implemented - I suspect it'll be applied very similar to AllTheWeb's clusters.

I've also seen a screenshot showing google displaying clusters at the top of SERPs - - I was not able to verify it as authentic, though. I'll try and track that down.

I, Brian · 10-31-2003, 05:01 AM

Found the link - here you go:

http://www.haymeadows.com/related.htm

Check the top of the page - "related searches"

burgi82 · 10-31-2003, 05:32 AM

Hi,
I don't really see that much evidence for the theme theory. The reason is that one of my competitors in the satellite tv industry has a link on every page of macworld (I think that is what it's called), a PR 8 site, and he ranks very well for several great keywords. He has a PR 7 because of these macworld links, and according to google, doesn't have that many other good links.
So, that example shows me that themed link partners don't matter too much, eventhough this case might just be an exception.
About crosslinking sites: there is this gambling webmaster who owns 5 gambling sites. He was able to get a good PR for his first gambling site, and now, has a good PR (7) for all 5 of his gambling sites. They are all crosslinked on every page of every website. He does have other links, but not that many on the newer ones of his gambling sites.
And in addition, all his sites rank number 1 for the targetted keywords, which means he is making a ton of money because the gambling industry pays (used to pay) very well. By the way, all his websites are hosted on the same server.
So, I have never done any study on the things I talked about above, but these are just observations that lead me to the conclusion that theme doesn't matter that much, and crosslinking your own network works also.

PhilC · 10-31-2003, 06:10 AM

Good find Brian. It's a pity that we can't get the URL for it. It shows that they are at least experimenting (or have experimented) with the idea of related links. Coming up with realted links doesn't mean that they will incorporate themes into rankings though, except when looking at the pages where the IBLs come from. In that way, I am sure that Google already 'themes'.

PhilC · 10-31-2003, 06:37 AM

I went looking in labs.google.com to see if it was there and I played with webquotes. I found 2 links to my site on their PR7 results page. The links don't appear in my pages' backlinks, so I submitted their URL to them. Weeeell.......you never know.

I, Brian · 10-31-2003, 07:40 AM

Quote:

Originally Posted by PhilC

Good find Brian. It's a pity that we can't get the URL for it. It shows that they are at least experimenting (or have experimented) with the idea of related links. Coming up with realted links doesn't mean that they will incorporate themes into rankings though, except when looking at the pages where the IBLs come from. In that way, I am sure that Google already 'themes'.

I could be quite confusing my terms - "related" with "themes" in particular. Either way, I'm looking to try and find how this may be further used for promoting sites where backlinks can be better controlled.

As for the graphic - in all honesty, I'm going to remain sceptical until there's an accessible link about - I'm sure it wouldn't be too hard for someone to create a misleading graphic.

Mel · 10-31-2003, 10:34 AM

Quote:

Originally Posted by I, Brian

...
At the end of the day, my argument is simply that almost randomly "link spamming" ie,

1/ posting in forums for the sake of posting only to get the sig link(s) up
2/ and also extensively cross-linking a network of domains -

are fairly pointless endeavours, and appear little rewarded.
....

I would have to see lots more evidence that Google has adopted Themes into their ranking algo before accepting such a theory. I have seen too many sites that rank very well on competitive terms as a result of anchor text back links from sites that have absolutely no common theme ( such as SEO links from Belgian travel sites for instance) but I do believe that you may be seeing the results of Google taking into account on topic text in the proximity of the link as Phil has mentioed.

As for links from an extensive cross linked network of domains, that is a tactic that works very well in Google, and accounts for the results that many commercial firms get for their clients. I believe that you might consider the Salsa network such a grouping and that is very successful.

**John Scott** · 10-31-2003, 11:17 AM

There are a couple cases to take into consideration. <edit>

Next would be the opposing case. At one time a web directory existed on www.bluefind.com - it had tens of thousands of links at one time. (All from my sites, on the same IP.)

BlueFind, with all those links, achieved a big fat PR1, and wasn't even #1 for the less than competitive KWD "bluefind", even though that was the link text used in all those links.

Just something to think about.

Mel · 10-31-2003, 11:37 PM

John, I agree that most of those sites which I have seen which are successful in operating groups of networked sites understand that it is not an excersise which cannot be undertaken without planning. I have seen such groups which use:

Fake owner names and addresses in the domian registration files
Hosting on different servers with unique IP addressess for all sites.
Different email addresses for different sites.

Do a Google search for engagement rings and take a look at the top ranked site for that term and in particular the secondary links at the bottom of the page which link to the other sites in the group.
Then take a look at the back links to that page and you will see that Google is crediting multiple links from all the sites in the group, plus links from the owners main site (which has been carefully left out of the group of links at the bottom of the page).

The links to this site from the pages of the other sites in the group are using the anchor text "engagement rings", a term that this site has achieved a #1 Google ranking for.

If you want to really research this you will find that a whois will reveal that all the sites are registered to different owners, but that many of the owners seem to have the same email address, and that the addressess are all within a 15 mile radius of a small town in Louisiana, and they all sell the same or similar products.

This is someone who knows how to make this system work.

You might also notice that all 14 of the sites in the link group have a batch of hidden text by way of a

[code:1:872d84b446]<DIV id=optimization
style="Z-INDEX: 500; LEFT: -231px; VISIBILITY: visible; WIDTH: 200px; POSITION: absolute; TOP: -130px; HEIGHT: 115px">[/code:1:872d84b446] though in this case I wonder why they have bothered as the hidden text is relevant to the page and site and would look fine visible.

burgi82 · 11-01-2003, 09:59 AM

Hi,
so, is this person using a different server and unique IP address for each of his websites, or does he only use different names for the registered user of the domains?
Also, what would you do if you want to link all your sites to each other? Unique host and IP for each site, different whois infos, or both?
Also, I saw a few examples lately where the hosting server, as well as the domain registration were the same for the complete network, and all websites of that person showed up in the backlinks. What sites? Make a search on google for either video poker, online casino, or online casinos. All of these sites, plus two more belong to that network.
do you think that google still bans sites that do this heavy crosslinking, or are they trying to find a way around this?
Would you do the crosslinking, and if yes, how would you do it? (see aboves question)
Thanks

PhilC · 11-01-2003, 10:06 AM

If you cross-link multiple websites in a heavy way (not just natural linking), then you must be prepared for the sites to be dropped. Google (and the others) do ban sites that they catch doing it, but they rely on people to tell them about it. If it's bad the sites will almost certainly be dropped. If it's not too bad, maybe they'll leave it alone but you can't rely on that.

If you are going to do it, it is best to seperate every site as much as possible, whcih includes all of the things you mentioned in your post. It won't guarantee anything, but it makes it a little harder to show that a group is connected.

burgi82 · 11-01-2003, 10:30 AM

Ok, thanks for the infos and warnings. But, what do you consider heavy crosslinking?
Ok, the way I do it right now is like done at my site at www.satellite-tv-4-you.com
See in the bottom, I have links to some other pages on the satellit tv domain, plus links to my other satellite tv sites. Acceptable, or heavy crosslinking?
Thanks for your advice and help!

PhilC · 11-01-2003, 10:40 AM

If all of your sites link to each other (even if they only link from one page) and are pretty much about the same things, then I'd call it heavy cross-linking. But if each site is about a distinct sub-topic, then I wouldn't.

For instance, let's say you have a main site about widgets. You could have another site that is all about blue widgets, and another one that is all about green widgets, etc., and the sub-topics wouldn't overlap much at all. Then they could 'naturally' link to each other. But if all the sites contain the same sort of stuff, even if they are designed differently, then it would be right to say that only one site is necessary and the others are just there for the linkpop/pagerank.

<added>
I'm not suggesting that you get rid of linkpop/pagerank sites. If that's what it takes to get to the top then do it. But you must be able to accept it if they all fall down.
</added>

burgi82 · 11-01-2003, 10:55 AM

Ok, thanks. I think that I know what I will have to do. I will pick the 4 sites that I want to be the big ones ( I have 8 right now on the same topic)
I will change the IP address for all of them, so they all have unique IP's. Then, I will not link to any of my other sites from the 4 big ones, except when I create other sites in the telecom sector.
But, on the other four sites, I will link to all of hte big ones, and also interlink them, but only so that they don't all link to each other, or better said, that not one link actually reciprocates.
Meaning, I will link from A to B, from B to C, from C to D, and from D to A, and with all these 4 sites, to all 4 of my big ones, giving each site a total of 5 links on them to sites in my other network.
I think that this should work, and I should not get any problems, since I don't do reciprocal links at all, but only one-way links.
Also, from what I see on other webpages that do use crosslinking and have all the sites on the same server, I should be fine.

Thanks for all your help!