The Location of Your Link on a Page Matters

[vangogh]

A link is a link is a link, right? All links are treated equally, right? A link in the main content of a page is the same as a link in the footer or sidebar, right?

The answer to all three is no. All links are not equal. As Google's PageRank algorithm has evolved it has become more sophisticated. Where at one time a link anywhere on the page would have been viewed the same as any other link on the page, today Google (and the other search engines) are better able to tell where on the page a link sits and they'll give more weight to a links located in certain parts of the page.

Here's a short video from Matt Cutts answering the question Are links in footers treated differently than paragraph links? (http://www.youtube.com/watch?v=D0fgh5RIHdE&feature=player_embedded) that I'd encourage you to watch. It'll basically tell you what I just said above, but this time it comes from Matt Cutts of Google instead of me.

The point to understand is that Google wants to see what they call editorial links. Links that are placed by a site owner because they believe the page on the other side of the link has value. Google would like to see a link as a vote for another page.

All links are not created equal. Some links are more equal than others.

[billbenson]

Hmm. guess I better rethink my linkfarm page :)

[vangogh]

Yep. You'll have to spend all weekend out in the link fields taking in the crops and planting seeds for next year.

[nealrm]

Given the example that Matt used, it sounds like Google MAY treat link that are repeated thoughout the site differently. So if you have a link in the same place on every page, that link could get less value. However, except for nav links, I don't see many links falling into this group.

[Spider]

Do you think it makes any difference to try to identify navigation links as such? -- perhaps with a remark ... <!-- NAVIGATION --> or <!-- MENU -->

How are local links and domain links treated differently, if at all? -- <a href="http://mydomain com/file.html"> as opposed to <a href="file.html">

And are one's own domain links treated differently from someone else's domain links? -- <a href="mydomain com/file.html"> as opposed to <a href="anotherdomain com/file.html">

More in keeping with this thread, what difference does it make if navigation links are spread across the top of the header or listed vertically in a lefthand column?

[billbenson]

A lot of cms's put heading tags in those columns. If you have those there, get rid of them. I don't think a comment tag will help you, but it won't hurt either IMO.

[vangogh]

Neal the odds are Google and the other engines will see all links outside of the main content differently than they will see links in the main content. Think of any part of your page that repeats across the site as part of the site template. Your header, footer, sidebar would all be part of the page less likely pass the same link value.

Frederick I don't know that you have to specifically mark parts of the page as navigation or menu. The search engines are working to figure this out on their own. Some would say you don't want to make it easier for them, because if anything it would mean you'd be helping them take away the value of some links. On the other hand html5 is adding some new tags along the lines of making it easier to identify parts of your page and things like microformats do the same.

There are advantages to marking parts of the page for what they are.

With the links the search engines are pretty good at figuring some things out. If you write a link simply as file.html they know it's really domain.com/file.html. There are things that seem to trip them up though that you'd think wouldn't. For example

domain.com
www.domain.com
domain.com/index.html
www.domain.com/index.html

will generally all point to the same place, but the search engines see those as 4 distinct URLs, because they don't have to point to the same place. If you're inconsistent in how you link to things each of the 4 URLs above could end up competing with each other hurting the overall ranking of the one page they all refer to.

Generally an external link (another domain linking to yours) will pass more value than an internal link (a page on your domain linking to another page on your domain). I say generally because it really depends on the sites and specific pages in question. Think of it though as what someone else says about you counting more than what you say about yourself. All other things being equal it's better to have a page on another site link to your page than having one of your own page link to that same page.

That's not to say internal links have no value. They do have value and sometimes internal links are enough to rank well for certain phrases.

As far as your links being across the top or down the side it shouldn't make any difference. Assuming a search engine can understand both are site navigation links they would likely treat them the same.

[billbenson]

domain.com
www.domain.com
domain.com/index.html
www.domain.com/index.html

This is something I have never completely understood. I understand what you are saying but I don't understand it if that makes sense :) I think its because I don't really understand how servers are configured.

In all cases they point at index.php. There are redirects set up by me or the server so no matter what you type, you end up on that page. If I type on my site http://domain.com I end up at http://www.domain.com. Had I typed in the path with index.php I end up at the same place.

So from a practical standpoint, I guess the server is doing a couple of redirects and if I redirect http.//www.domain.com/index.php to http://www.domain.com I should be fine.

If the server isn't redirecting http://domain.com to http://www.domain.com I should do that?

Some servers have you put your site under www. For my host, I just put everything under public_html. There is no www. In the first case I can see that the www is a subdirectory. In my case I'm a subdirectory of public_html.

I think I just confused myself writing this... Whats going on and what should be done so G never sees these as different pages?

[Bill Slawski]

Hi Bill


Whats going on and what should be done so G never sees these as different pages?

It's possible to use many different subdomains at a domain, and even set things up so that the "www" version shows different content than the non-www version. These days, most hosts set a site up by default so that the "www" version and the non-www version both resolve to the same place. Search engines will often treat both as if they are different pages, since they use different URLs.

The ideal solution has a few parts:

1) Choose one version or the other (www or non-www), and set up a 301 redirect to point all URLs for one version to the other.

2) With all internal links on your site, consistently link only to the version that you have chosen, whether it's the www or the non-www version.

3) In Google's Webmaster Tools, in the Settings area, choose the domain preference that you are following, whether the www or the non-www version.

4) When submitting the site to directories, using it on profile pages off site, and in other places where you can place a link to your pages, always use the version that you have chosen.


So from a practical standpoint, I guess the server is doing a couple of redirects and if I redirect http.//www.domain.com/index.php to http://www.domain.com I should be fine.

The search engines also don't know the setup of your server either, and they can't tell that the default page for your root directory is index.php, or index.htm, or default.shtml or home.aspx, or whatever else it might be.

It's not a bad practice to link directly to your home page domain if possible in links from your logo and any other links to your home page, without including the "index.php," or whatever may indeed be the default file name for your home page.

A 301 redirect to the version of your domain home, including a trailing slash from that default file name can be a good idea, especially if sometime in the future, you want to change from using php to html or aspx or shtml or whatever file extension you may choose.

But, make sure that you actually change the URLs in your links on your site to point directly to that version of the URL, so that you don't send visitors to the pages of your site through a redirect every time they go through a link to your home page. The 301 redirect should be for visitors coming to your site from outside of your pages, whether they are typing your address into their browser, or visiting from a link that includes the "index.php." That includes visits from search engines as well.

[Spider]

Fascinating! Thanks for all your comments. I'm actually understanding about half of this!

FWIW, I have for some time been using non-www domain names for all linking except when sending my link in e-mail. My business cards also carry the www version. Although I'm sure it's not especially accurate, my stats separate the clicks from e-mail and entry from business cards (www domain) from the clicks off the web (domain.)

[vangogh]

Thanks for dropping by Bill and for the added explanation.

Frederick I just took a look at your site and noticed I could access your home page with and without the www and with and without the index.html. If you go to any variation you'll notice you can see the URL you actually typed. What you want is to pick one and redirect the other versions to that URL.

Since most of your links use the non www version you should probably redirect things to frederickpearce.com. Then if you type LIFE COACH, Business Coach and Personal mentor - Frederick Pearce, Houston, Texas (http://www.frederickpearce.com) the URL you end up seeing in the address bar would show as frederickpearce.com without the www. People could type or even link to any of the variations, but once at the page they would only see the one version of the URL.

Here's code you can add to your .htaccess file to make it happen



RewriteEngine On
RewriteCond &#37;{HTTP_HOST} ^www\.frederickpearce\.com [NC]
RewriteRule (.*) http://frederickpearce.com/$1 [R=301,L]

redirect 301 /index.html http://frederickpearce.com


I'm pretty sure your server is Apache and you can add an .htacess file. When creating the .htaccess file remember to use a text editor like Notepad and not something like Word. Also the file has no name, just the .htaccess extension and it should be located in your root directory (the same place where index.html is located).

It's possible you already have an .htaccess file on your site. In that case you would add the above code to it, but probably don't need the first line turning the rewrite engine on.

[KristineS]

This is great info. Need to get my brain a bit more online to assimilate it, but very useful.

[billbenson]

Thanks Bill. Great explanation.

Is this then a reason to use absolute links over relative links on your site?

[vangogh]

You can use relative links and search engines will be able to fill in the rest. The advantage to relative links is that it's easier to move the site later as long as you keep the same directory structure.

I do prefer to use absolute links though, for a couple of reasons.

If someone steals your content the absolute link still points back to your site. It still sucks to have your content stolen, but at least you get a free link out of it.

If your content is publishes via rss or gets emailed then you need to have the absolute link or it won't work. A relative link in that case would be relative to the rss feed or email, which wouldn't know about your directory structure.

Bill might have more to add, but as far as I know search engines will see either an absolute or relative link the same way as long as they point to the same place. Still I prefer using absolute URLs for the reasons above and the one negative of absolutes making a site harder to move around, usually isn't all that difficult to work around.

[vangogh]

Found another post on the subject, SEO Terminology: Boilerplate (How Google May be Treating Repeated Content) (http://www.searchenginejournal.com/boilerplate-how-google-treats-repeated-content/12682/). Boilerplate is just another term for your site's template or the stuff that repeats across pages on your site.

[Spider]

Thanks, VG. I'll check it out and see if I can do it.

[vangogh]

As soon as you create and save the .htaccess file on your site check the site in a browser and make sure it's doing what it's supposed to. If anything goes wrong, just remove the code or the entire .htaccess file and let me know. It's easy to make errors with the file, but none of the changes are permanent. As soon as you remove the .htaccess file everything will go back as it was.

[Bill Slawski]

I like using absolute links too, but it can be convenient to use relative links during development. If done right, either way is fine with search engines.

As for the topic of this thread, one of the early papers from a major search engine on the topic of breaking a page down into visual segments or blocks was from Microsoft. You can find it at:


VIPS: a Vision-based Page Segmentation Algorithm (pdf)
http://research.microsoft.com/en-us/um/people/jrwen/jrwen_files/publications/vips_technical%20report.pdf

The basic idea is that a page can be broken down into different parts, and a search engine might be able to identify those parts by looking at the HTML code for the page, and by actually looking at whitespace, horizontal rules, and other layout features that separate parts of pages.

So, a page might be seen as having a heading area, a footer area, a main content area, a main navigation section, a sidebar or two.

Microsoft researchers followed up with a number of other papers and patents where they discussed things such as what the most important picture might be on a page (probably in a main content area, and probably the largest picture, amongst other considerations). Some pages have multiple important blocks, such as a newspaper site that might have a number of different stories on a front page.

They also discuss how they might treat links differently when they appear in different segments. A Microsoft paper that discusses this is:

Block-level Link Analysis (pdf)
http://research.microsoft.com/en-us/um/people/jrwen/jrwen_files/publications/block-level%20link%20analysis.pdf

When I did a recent site search for my site at Bing, I noticed that Bing was including search results for individual comments from my blog as separate entries. Kind of interesting to see.

Now that's Microsoft, but Yahoo and Google have also written about segmenting pages into different parts, and there have been some comments from people at those search engines who have intimated that links are given different weight based upon where they are located on a page. Matt Cutts said something like that in a video within the part couple of months.

We also heard something similar in an interview with Yahoo's Priyank Garg ( Yahoo's Priyank Garg Chats with Eric Enge (http://www.stonetemple.com/articles/interview-priyank-garg.shtml) ):


Eric Enge: Right. So part of what you are pointing at there is that relevance matters a lot. So getting a link from the bottom of a WordPress template that you create and distribute is completely irrelevant.

Priyank Garg: Exactly, that’s the kind of thing that we are trying to do all the time. The irrelevant links at the bottom of a page, which will not be as valuable for a user, don’t add to the quality of the user experience, so we don’t account for those in our ranking. All of those links might still be useful for crawl discovery, but they won’t support the ranking...

[vangogh]

Bill sometimes I'll use root relative URLs in development so at least if they don't get changed to absolute URLs later I know they'll still work. As you do more with includes or develop on one of the popular CMSs it becomes easier to just use absolute URLs. They all tend to be in one or two files and are pretty easy to change.

It can become a pain trying to change links in your content if you later have to move the site, but I tend not to build the content pages on my development server. Once the template or theme is done I usually move it online if I can.

Page segmentation and block level analysis makes a lot of sense to me. I realize some won't like it as they probably have tons of links in those parts of the page likely to carry less weight, but when you consider what search engines want links to be it makes a lot of sense to give different weight to links on different parts of the page.

html5 and microformats are going to help them break down the page. It won't surprise me if some sites see big ranking changes once html5 becomes the de facto standard most of us are using.

[Bill Slawski]

Steve,

I tend to like to use absolute URLs as well, but many of the sites that I work on with clients are created or updated on development servers and remain as relative URLs when they get moved onto production servers.

It does make a lot of sense to give more weight for ranking purposes to links that appear in main content areas of a page than in boilerplate or navigation areas, and the ideas have been around for a while. The VIPS paper came out in 2003, and the block-level link analysis paper in 2004, so the idea that links may be given different amounts of weight when they appear in different areas of a page has been floating around for a while.

One section of the block level link analysis paper discusses what they call block level pagerank:


4.1 Block Level PageRank

Block Level PageRank (BLPR) is similar to the original PageRank algorithm in spirit. The key difference between them is that, traditional PageRank algorithm models web structure in the page level while BLPR models web structure in the block level.

I think that's important because it shows that search engines aren't necessarily tied to assigning pagerank to individual pages as they may have been in the past.

HTML 5 may be a step that can help search engines in understanding different parts of pages better, though there will likely still be a visual segmentation approach used that analyzes where things appear on a page in addition to the HTML code on that page.

Microsoft followed up the papers on Visual Segmentation with something that they called Object-Level ranking. See:

Object Level Ranking: Bringing Order to Web Objects (http://research.microsoft.com/en-us/um/people/znie/f611-nie.pdf) (pdf)

The use of microformats on web pages can help search engines extract facts from pages more easily, and may help in that kind of Object Level Ranking.

[billbenson]

<?php echo $_SERVER['SERVER_NAME'] . '/rest-of-url.php'; ?> should get you a transferable absolute URL as well?

[vangogh]

@Bill B - yep, but that does assume you have PHP running locally. I do, but not everyone, even all web developers, do. If there are enough links that would need to be changed I probably would go the route you suggest.

@Bill S - I think we develop in a similar way. Thanks for the additional PDFs. I think I've read the block level PDF and most likely grabbed it via one of your posts (or one of Dave's) I don't think I've read the Object Level PDF so I'll add that to the list.

The whole thing makes so much sense to me. It's not hard to see how a link in the footer or in a blogroll isn't quite the same as a link inside the content of the page. On the other hand it doesn't necessarily mean the link in the boilerplate should be the one with less weight. If I link to you in my blogroll it's really more of a vote for your site than a single link to a page in a post I write. The two links should be treated differently, but one isn't (or shouldn't be) automatically better than the other based solely on page location.

I agree that search engines will look at more than semantic html5 tags. The new tags will help them, though for the most part we tend to name divs consistently now. I can also see how in the future some will experiment with different tags to see which if any leads to better ranking. I can already see the abuse.

[Bill Slawski]

On the other hand it doesn't necessarily mean the link in the boilerplate should be the one with less weight. If I link to you in my blogroll it's really more of a vote for your site than a single link to a page in a post I write. The two links should be treated differently, but one isn't (or shouldn't be) automatically better than the other based solely on page location.

One of the reasons why I spend so much time looking at patent filings and white papers is to see some of the assumptions that search engineers might be making, as well as possible processes that may or may not be implemented by them. The assumption that a link in the main content area of a page might be more important than one in a sidebar may be something that we disagree with, but it's possible that the people coming up with potential algorithms that may be used on search engines feel differently.


The new tags will help them, though for the most part we tend to name divs consistently now.

I think that can help. Knowing that a search engine might not give as much weight to content and links (and images) and other things, in certain parts of a page, and trying to label those parts in a manner that the search engines might recognize means that you have a better understanding of how a search engine might handle the content that you are creating.

[vangogh]

One of the reasons why I spend so much time looking at patent filings and white papers is to see some of the assumptions that search engineers might be making, as well as possible processes that may or may not be implemented by them.

Which is one of the reasons your blog is at the top of my list of feeds. :)

I've always tried to understand what the search engines might be thinking and where they might be going and in general understanding how they'd like things to work. I never wanted to be someone who chased the latest tactic only to have to change again in a couple of months. I prefer to work at things I expect will last.


The assumption that a link in the main content area of a page might be more important than one in a sidebar may be something that we disagree with, but it's possible that the people coming up with potential algorithms that may be used on search engines feel differently.

True. This is probably one of those places where search engines are looking more to eliminate spam than anything else. And in general a link inside your content is probably more valuable than one in the footer. Still I can see how at times the link in the boilerplate is really closer to the "vote" Google might want to see.

The developer in me is naturally interested in html5 and what the new tags will bring. We're still some time away from it being used in practice and then perhaps it'll be a little longer before seeing how search engines begin treating them. Should be interesting and I'm sure you can already see the "Top html5 tags you have to use to improve your seo" posts coming.

[yummytork]

And if any of you's guys are wondering 'where in the world can I get all these wonderful, glorious, editorial links' then you might want to check out linkvana. We've been testing it for a few months now, and it works like a charm. Their reporting tool sucks though (they don't have one)

[vangogh]

Linkvana seems like the typical automated aid link building. A little different sure, but in the end the kind of links search engines don't want to give weight to. Maybe it will work today, but no reason to think it will continue to work. In the end it's another attempt to manipulate rankings so it's the kind of thing search engines will work to counter. They might not be able to detect the links as being bought and artificial today, but long term I would think they'll figure it out.

Not saying you shouldn't use the service, but know you're probably buying something temporary.