The Ultimate Guide To Internal Linking

Posted by Pete - 07/08/08 at 04:08:38 pm


There are a lot of ways to approach the internal linking of a site. Actually, that’s a bit of an understatement. There’s hundreds. So I thought I’d write a guide on how to design and impliment internal linking that’s elegant and works, so that both your readers and the spiders alike will love you forever.

There’s also an issues part below that, where we’ll deal with how to correctly counter issues with pagination, navbars and so on.

  1. How PageRank Works
  2. Planning The Structure
  3. Dealing With Pagination
  4. Navbars and Sitemaps
  5. What Not To Do

How PageRank Works

Firstly, we’re going to have a look at the original PageRank formula, as it’ll give you an idea as to what’s going on when we tweak things later. Please bear in mind that toolbar PR is a load of rubbish, and that the equation has changed since this came out. That said, it’s still a fair indicator, so it’s worth knowing.

The PageRank formula, at its simplest, is this:

PR(pi) = (1-d/N) + d * (Σ pj ε M(pi)) * (PR(pj)/L(pj))

…where p1,p2,…pN are the pages in the site, M(pi) is the group of pages that link to pi, L(pj) is the number of outbound links on page pj (where pj is a page that links to pi), and N is the total number of pages in M(pi).

However, if you’re human and normal, I’m guessing that won’t mean much to you, so think of it this way:

The PR of a page is determined by the amount of PR passed to it by all the pages linking to it. The PR passed by those pages is determined by their PR, divided by the number of outbound links on them.

So imagine we have a page. We’ll call it Phil. And 10 pages link to Phil, each with a PR of 5. Each of those pages has 5 links on it. So every link on one of those pages passes 1 unit of PR. Thus Phil has a PR value 10. N.B. this is not the same as toolbar PR. This is the actual value, not the assigned level.

Planning The Structure

Firstly, let me qualify this… This method of planning will work for pretty much any site, as long as you follow every step properly. This breaks down into three steps:

  1. Filtering
  2. Prioritising
  3. Dividing

Filtering

The first step, filtering, deals with getting the stuff you don’t want indexed out of the equation. Here’s a quick list of pages you can nofollow all links to:

  • Privacy Policies
  • Accessibility Policies
  • Legal Terms

…and stuff you can remove in robots.txt

  • Shopping carts
  • Search results
  • Thank you pages
  • 404 pages

You may also have site-specific things that you want removed too. As a general rule, work on the principle that if it’s not useful to someone coming straight to it from a SERP, it probably shouldn’t be linked to with any link that’ll pass PR.

Dividing

Now you’ve got to sort out where you want to divide the site. Services/products etc should be bundled together, top level boring pages, like Contact and About pages would go in another ’section’ and so on. Work out which bits go with what, and then meet the final planning step…

Prioritising

Finally, you need to decide which of your divided sections are the most important, and what out of those areas are the most important pages. So if for instance, we have a site selling a variety of products, the homepage would probably be 10, the individual product pages would be about 5, the top pages for product selections would be 7/8 and anything else would sit at around 2 or 3.

This will not only help when you’re sorting out the flow of link weight through the site later, but also when you come to make an XML sitemap, as the priorities should match the link flow.

Dealing With Pagination

If you’re running an e-commerce site, chances are you’ve got pagination issues somewhere. If you don’t know what pagination is, it’s the consecutive numbering you see on the bottom of the pages of some sites. The number links at the bottom of a Google SERP are a good example.

Now, the problem with pagination is that it leaks link weight like you wouldn’t believe. However, it’s pretty necessary for humans. So, I use the same method to get around it as outlined by Rand in the Give It Up Whiteboard Friday. What you do is you nofollow all the paginated links, and all links to funny sections and drilldowns. You don’t want to exclude them in robots.txt, but just stop them passing link weight.

Now, create a set of pages with links to the pages that we want ranked. So if you’ve got 100 products, create one or pages that are a list of links to all those products. Then link to those link-pages via the footer. This gives you a great way to serve people and spiders properly. If you want to be really clever as well, you can put in some javascript to detect if someone’s coming in from a SERP, and then re-direct them to the actual normal user page. However, this is borderline blackhat, as it’s technically cloaking, so beware.

Anyway, whether you cloak or not, you now have a set of pages (the normal site pages) that work for people, and a set that are search-spider-tastic. For a slightly better explanation, check out the Whiteboard Friday video. There’s lots of other good stuff in it too.

The problem with navbars and footers is this: they tend to have links on them that people want, but that we don’t want to lose link weight to. So how do we fix it? Well, nofollow all the links to pages that aren’t as important, same as you would in a page’s copy. That way all those links to Contact Us or About won’t leech away your internal linking awesomeness. The same goes for sitemaps.

Basically, every link that you put it, ask yourself “does that page/section need more link weight?”, and if the answer is no, nofollow it. Don’t go overboard though, or else you won’t have any internal links as far as the engines are concerned.

What Not To Do

A few things to beware of:

  1. Don’t over-link.

    2. I’m looking at some of the big blogs here. Engadget do it, and it pisses me off no end. If you’re going to put in a link with the anchor text “AMD”, make it link to AMD’s site, not your page on AMD. Make links to what the user would expect them to.

  2. Don’t go mad with nofollow.

    3. Please leave something that the spiders can see…

  3. XML sitemaps can screw you.

    4. Make sure you set your priorities right. If you don’t, it’ll come back to bite you in the ass.

  4. Human’s first.

    5. Yes, it’s important that you have a site that’s search friend from an internal linking perspective. But search engines don’t buy things, people do. So make sure you please your human users first.



    Get a Trackback link

    5 Trackbacks/Pingbacks

    1. Pingback: Guide To Internal Linking on August 8, 2008
    2. Pingback: Link Building this Week (32.2008) | Wiep.net on August 8, 2008
    3. Pingback: Why your SEO advice sucks so bad | SEO Theory - SEO Theory and Analysis Blog on August 9, 2008
    4. Pingback: Why YOU Shouldn’t Be Using Social Media :: Searchlight Digital on January 2, 2009
    5. Pingback: Best Internet Marketing Posts of 2008: Social Media, SEO, and More ยป techipedia | tamar weinberg on January 5, 2009

    Leave a comment

    You must be logged in to post a comment.