| 9:34 pm on Mar 14, 2008 (gmt 0)|
Since I just welcomed you in the Apache forum [webmasterworld.com] I noticed you had this old post out here. It has been two years later, what have you discovered? :)
| 7:14 am on Mar 15, 2008 (gmt 0)|
Google certainly doesn't ignore them, but IMO they aren't worth as much as links from a completely different IP block.
Plus, if you're going to do this, think about the arrangement (I mean draw a diagram). I've seen proof posted elsewhere that networks of sites still work in Google, but don't get greedy! Heavy crosslinking between unrelated domains has to be a bad idea long term. If I were Google, I would have a limit to the number of inter-site linking I would tolerate on one IP block.
| 9:20 am on Mar 15, 2008 (gmt 0)|
I always use different servers wouldn't dare do otherwise. Sounds like your just going to make 5 low quality make do sites. I think you should save your time and just buy some quality links (be careful of course).
| 12:07 pm on Mar 15, 2008 (gmt 0)|
just do it one way links. links from the 5 to the 1 do not link from the 1 to the 5.
If you are still scared just link without the anchor text and simply pass PR without context.
| 3:42 pm on Mar 15, 2008 (gmt 0)|
The overall health of the backlink profiles involved will make a big difference here, too. This was not as obvious 2 years ago, when this question was first posted, but it has become a lot clearer this past year.
[edited by: tedster at 6:48 pm (utc) on Mar. 15, 2008]
| 6:15 pm on Mar 15, 2008 (gmt 0)|
Also, don't use site-wide links on the 5 linking to the 1. 100 links from one site will help no more than 1 if they are all identical. I've read that by varying the anchor text you can get Google to count more, but again consider you're possibly increasing risk. If the sites are related then the best place for the links is in body text content. Even if they're not, you might be able to find a way to add some content that relates to the site you're linking to and link from within that article / page.
| 9:12 pm on Mar 15, 2008 (gmt 0)|
"100 links from one site will help no more than 1 if they are all identical."
Of course they will, they just won't be 100x better.
| 11:25 am on Mar 16, 2008 (gmt 0)|
Have you tested this yourself? I have. Identical site-wide links (for instance in a site template or footer) ARE NOT counted beyond the first.
About 18 months ago, before we started promoting our design site, the only IBLs it had were from 60-70 of our clients websites. These links all used to be part of each site's template, however we changed them all over the course of a fortnight so our link was only on the home page of each site. No change in our positions in Google for any terms.
| 5:40 pm on Mar 16, 2008 (gmt 0)|
One could even argue... and I haven't tested this, but I think I've seen it in action... that sitewides, by their nature, go beyond editorial comment and suggest a special relationship between sites to Google. As such, it's possible that they may ultimately hurt you (depending on the rest of your link profile)... by making the sitewides worth actually less than a single link.
As an advertising strategy on a very large site where no one page gets a lot of traffic, they may well be a good choice... but they are most likely advertising and probably do raise a flag.
[edited by: Robert_Charlton at 5:49 pm (utc) on Mar. 16, 2008]
| 12:36 am on Mar 17, 2008 (gmt 0)|
"Identical site-wide links (for instance in a site template or footer) ARE NOT counted beyond the first."
Your contention is beyond absurd. The idea that a site gets the same benefit from one link in dmoz is the exact same benefit as Wikipedia or opensite gets from sitewide dmoz links is utterly ludicrous.
Of course sitewide links are still one of the most powerful things to benefit websites, much much moreso now since Google's crawl is so much weaker than previously and pagerank counts for less. One PR4 link to a page will leave it in the supplemental index. One PR4 link, plus a mix of 1000 PR3, PR2 and PR1 links from one site will keep that page in the main index, and get it to rank for the link text of those links.
There may be no more FUDdy thing around these days than the misinformation involving sitewide or blogroll links. Just because 100 of them aren't 100x better than one doesn't mean they are bad, and it is plainly obvious there is great benefit from multiple links from the same domain, especially from high quality sites.
Spam makes things of negative value, but its the height of FUD silliness to say it hurts the Wikipedia to have all those DMOZ links.
| 1:24 am on Mar 17, 2008 (gmt 0)|
steveb - I'd make a distinction between multiple links from the same domain and templated sitewide links.
| 7:41 am on Mar 17, 2008 (gmt 0)|
That's irrelevant to the issue. So if 100 multiple links are worth 57x, and templated links are worth 36x, that is still far more than 1x.
The dmoz wikipedia example shows there is no downside to templated sitewide links. Zilch. And in fact there is a very, very powerful benefit to get such links from excellent sites. Of course sitewide links from a piece of garbage site could hurt you.
| 12:25 pm on Mar 17, 2008 (gmt 0)|
Getting back to the main issue, about links from sites on the same host & IP...
|I dont want to have 5 seperate hosting accounts for all my domains, so im going to do an 'add on domain' all on the same server. my question is this: does the google spider know that these 5 domains are on the same host, and thus ignore them, or does it see them as 5 completely seperate sites? |
How about the reverse? How about if there are outbound links from a page on a site to 5 other sites that are all on the same host/IP?
| 6:39 pm on Mar 17, 2008 (gmt 0)|
|How about if there are outbound links from a page on a site to 5 other sites that are all on the same host/IP? |
If this linking pattern forms a large enough percentage of the inbound links to the 5 other sites, I've observed that the sites will have problems on Google. It depends very much on how competitive the searches are. The pattern, though, is too unlikely to be natural, and Google can easily spot it.
| 7:42 pm on Mar 17, 2008 (gmt 0)|
who said it must be sitewide links? why not make a script to vary that so that you can get multiple anchor texts working.
| 8:15 pm on Mar 17, 2008 (gmt 0)|
steveb: you didn't answer this before, so I ask you again - have you tested this yourself?
I repeat, I have, with 60 odd sites (PR1s,2s,3s,4s and a few 5s), nearly all with no other OBLS, linking a site that had no other IBLs - and I know what I saw. No drop in rankings. Therefore the anchor text benefits of identical site wides are ignored beyond the first link, whatever you might prefer to think. I shared these results with others on another forum at the time and a few had conducted their own tests and concluded the same.
As far as PR goes, toolbar was constant at a mix of 5 and 4 - the true value may have fallen. If there is additional PR from extra sitewides then I would say the amount passed is *seriously* devalued if it is counted at all.
|Of course sitewide links are still one of the most powerful things to benefit websites |
sitewide INTERNAL links = very useful for showing Google your most important pages.
sitewide EXTERNAL links = much abused. Matt Cutts (when he does talk about site specifics, which is rare) has commented on non-related site-wides as being a low quality indicator.
|The idea that a site gets the same benefit from one link in dmoz is the exact same benefit as Wikipedia or opensite gets from sitewide dmoz links is utterly ludicrous. |
First, DMOZ is a bit of a bad example because for all its faults Google appear to treat it as a holy cow. Wikipedia has millions of links - how do you *KNOW* that those sitewide DMOZ links are being counted?
|height of FUD silliness to say it hurts the Wikipedia to have all those DMOZ links |
I never said sitewides would hurt, I just said they wouldn't count.
|There may be no more FUDdy thing around these days than the misinformation involving sitewide or blogroll links |
Well, do you have specific information based on observations to share? Or just a hunch? To my mind, a 'ridiculous', 'beyond absurd' 'FUD' contention would be one with no facts to back it up.
| 8:31 pm on Mar 17, 2008 (gmt 0)|
|How about if there are outbound links from a page on a site to 5 other sites that are all on the same host/IP? |
|The pattern, though, is too unlikely to be natural, and Google can easily spot it. |
But when the pattern is natural, Google does not seem to have issue. Especially when local. Take for instance community sites. Often times entire communities and their sites will be hosted on the same server. The base community site itself will have a directory listing of local area businesses and offer links to those businesses/organizations as well as mention of them (and links to them) in news articles, calendar of events, etc.
| 7:01 am on Mar 19, 2008 (gmt 0)|
|But when the pattern is natural, Google does not seem to have issue. Especially when local. |
I agree, and local is one of the things I had in mind when I said: "It depends very much on how competitive the searches are." Google generally doesn't have issues with material that is very unique. As the searches become more competitive, Google begins to have issues about more and more things.
So, ranking for...
small town keyword widgets
...is going to be easier than ranking for...
big city keyword widgets
...which is going to be easier than ranking for just plain...
In the more competitive searches, Google appears to care more and more about where you get your links from.
Whether Google makes specific adjustments for sites all served by the same local hosting company is an interesting question. They probably don't need to.
| 10:35 pm on Mar 19, 2008 (gmt 0)|
"have you tested this yourself?"
Obviously, and also just as obviously the effect can been seen on dozens of my competitors, and obvious sites like Open Site and Wikipedia.
Seriously, it's just absurd these days to say that 1000 PR4 links matter the same as one PR4 link. In the latter case the page will almost always be supplemental and rank for nothing.
| 11:51 am on Mar 20, 2008 (gmt 0)|
If you have tested, then would you share the particulars?
Perhaps links between different IPs are treated differently.
All of the sites involved in my test were on the same IP (which would still make my advice highly relevant as that is the situation here and the topic of the thread).
What did you do and what did you notice? How many sites? How many pages?
| 9:33 pm on Mar 20, 2008 (gmt 0)|
Every site every time every page. Going from one link to five links will boost a page.
Additionally your statement really is extremely outlandish when looked at in a micro fashion. A link from Dmoz to a Russian page will help. A second link from a Dmoz page to a Polish page will show a near immediate boost to the Polish page and no drop for the Russian page, in fact a small boost will occur.
Getting 100 links from domainA to one page on domainB is one thing. Getting 100 links form domainA to 75 different pages on domianB is another. The added boost from the 99 redundant links is good, but the boost to the 75 different pages, and thus the whole domain, will be far more significant than a single link to one page.
(Fianlly, if you think you give the same benefit to a page by internally linking to it from one page as you do from 1000, I don't know what to say. Google similarly diminishes internal links, but certainly does not discard the value of all links besides the first one, and that is super easy to test.)
The test you are talking about seems mostly not relevant because the original and destination sites are on the same IP, which means they all could be penalized or categorized already as an inbred neighborhood.
| 10:30 am on Mar 21, 2008 (gmt 0)|
By 'site wide' I mean *identical* external links - same anchor text, same position. NO anchor text boost for the additional links because when they were removed no drop in rankings occured. This tested in isolation.
|test you are talking about seems mostly not relevant because the original and destination sites are on the same IP, which means they all could be penalized or categorized |
They are being counted. As the OP is talking about an identical situation my original advice to him is bang on the money. I did mention that varying position and anchor text could get Google to count more than one link.
|Fianlly, if you think you give the same benefit to a page by internally linking to it from one page as you do from 1000, I don't know what to say |
Neither do I, but if you read my posts, I never said sitewide *internal* links are disregarded. I said they were very useful.
I have the ability now to set up a test from completely different IPs, which I will do.
| 11:04 am on Mar 21, 2008 (gmt 0)|
It's odd to see someone assert that suddenly internal links are better than external ones. Your suggestion is contradicted not just by all evidence and easy testing, but by the behavoir of zillions of webmasters. Many blogs rank well based strictly on blogroll links from three or four domains. They do not rank well based on ONLY three or four links and three or four incidences of anchor text.
Also, original poster did not say all links were from the same IP to the same IP, only that all links were on the same IP. S/he did not say what the IP of the destination was.
Anyway, you have your opinion, and I voiced mine, so everybody else can decide what they want.
| 12:45 pm on Mar 21, 2008 (gmt 0)|
|Since I just welcomed you in the Apache forum I noticed you had this old post out here. It has been two years later, what have you discovered? |
My guess is he discovered responses to new topics are easier to come by these days... He got two in less than 24 hours from his second question.
(The first one didn't go so well.)
| 3:39 pm on Mar 21, 2008 (gmt 0)|
I've also seen some websites develop ranking troubles who offer cross domain (and cross-IP) links from their main menu. That's apretty extreme run-of-site situation. I'm not yet sure if this is straight cause and effect or just a coincidence - but I thought this thread was close enough to the topic to make a mention.