Nofollow & PageRank Manipulation

June 17th, 2009 by Chris

Matt Cutts made a blog post you really should read.

I’ve been talking about PageRank manipulation for a long long time now, almost 8 years, going back to my time on SitePoint, doing experiments, getting a page a PR 7 with a single incoming link, editing menus to siphon PageRank to pages I desire (such as in my Hub & Spoke article). I’ve probably advocated it as a practice longer than pretty much anyone else (when I started other “professionals” often said I was full of it (mostly because they didn’t understand the math behind PageRank), only to start doing it themselves a few years later, many of them not until apparently 2007, 5 or 6 years late), the most being written in my article on internal site architecture a phrase I started using before anyone else (forgive the horn tooting, but I’m not a sensationalist writer, and I feel that the only way I get readership on this site is because on many SEO topics I was first, and I was right. So I just like to reiterate that while other more popular SEO writers are indeed more popular, they have a crappy track record). So, if you read my blog, and you’ve read my stuff, you’ve probably adopted such practices yourself, and so, you really need to read that blog post.

There are a couple of good points in that blog post. For instance Matt Cutts acknowledges what he calls “PageRank Sculpting” as a legitimate method of SEO, though he says it is something that he would consider to be “third tier” and that you should work on having good titles and content first. He also mentions menu control as a valid practice, controlling which pages are linked on your menus or from your homepage, something I taught 8 years ago (hooray vindication).

The main topic though is about rel=nofollow. Typically going back years the goal of limiting the outflow of PageRank to sites (thus keeping more within your site, and again something other SEO writers claimed was impossible, or not beneficial, or that you didn’t even lose PageRank when linking out..) was done with form based navigation, javascript based navigation, blocking links through a relay script, or all of the above.

These were all highly technical methods of blocking and more difficult to implement. Then with the advent of rel=nofollow we all had a simple and easy way to block the transfer of PageRank. Until apparently 1 year ago. 1 year ago Google, secretly, changed how they calculated PageRank (which again, this whole conversation is an implicit statement that outgoing links do indeed take PageRank away from a site’s pool in general, something that had been a contentious issue for years, as anyone who visited a forum can attest to). The change was that, in Matt’s words, instead of dividing the value a page has by the legitimate links on a page to see how much weight each link had. Instead all links on the page are now considered, ie both normal and nofollow links. So if you have 10 links on a page, and 5 are nofollowed, the total link weight is divided by 10 not 5, and the PageRank that would have been passed to the 5 nofollow links simply evaporates, rather than being passed to the 5 normal links on the page.

This is a huge issue. Suppose you have a blog with a menu of internal links totaling 20 links, and your blog post has 100 comments because you’re just that popular, and suppose 60 of those comments have outgoing links, but you rel=nofollow them because they are user contributed links and you cannot vouch for them (they could be “bad neighborhoods” and all that). Prior to this change any PageRank passed from your site, or from external sites, to this blog post would have been funneled back through your 20 internal links to the rest of your site. Your net PageRank loss, or bleed rate, would have been small, equivalent to the part of the PageRank formula known as the dampening factor.

Now, 75% of your weight is lost because the 60 nofollow links are included in the equation. That is huge. This could be why, it seems to me, there was a global PR reduction about a year ago (many sites lost PR, though relative rankings stayed level). All that PageRank evaporating instead of continuing to go around the circle. This doesn’t even just affect you if you run a blog, but also if you otherwise have links from blog posts, those links are now passing less weight to you.

Additionally, what about using rel=nofollow on your own internal links as has been widely suggested prior to this change? Well, suppose you used rel=nofollow on a login page because search engine spiders don’t need to login to your site. What exists on that login page? Typically your header, your side menu, and your footer. Previously it was suggested to nofollow such links, to prevent unnecessary drain. This is no longer a good idea. If you nofollow an internal link you lose ALL weight that link would get, whereas if you kept it live you’d lose some to the dampening factor, but much of it would flow back through the menus on that page. Better to lose only some of it than all of it. This goes for forums, and shopping carts, and any other user software with copious amounts of useless (to search engines) pages that you would have otherwise nofollowed. [b]From now on, never nofollow an internal link.[/b]

No word yet if this also applies to pages blocked with robots.txt or a noindex meta tag. Many people, including myself, use those two methods to block unwanted pages, especially in regards to forums. Forum software trots out so many different useless URLs this is a big deal. For instance at any given time on a popular forum, say SitePoint for example, you’ll have thousands of links on your homepage pointing to things like user profiles. You don’t however have thousands of forums and the user profile links drastically outnumber the actually forum content links, so you block the user profiles with robots.txt or noindex and ideally send all the link weight to your content. If this change applies to noindex and robots.txt, you’re instead just sending over half of your available link weight into the abyss. Bad idea.

So, if you still want to do PageRank Manipulation (and you should, it is the one thing that separates professionals from amateurs and can give you good success and the edge you need) these are the things you can do.

1. Control your homepage links and your menu. On my sword site I customize my left menu by hand. The software by default would link to the top level subcategories. I do that, but also pull out some products individually for direct links. This is to funnel more PR to those product pages to help them rank better. I do this because I want those products to rank better because a good rank on those products gives more profit because the product is more often searched for.

2. Use strength in numbers, if you can have big internal link menus you will keep more of the PageRank within your site regardless of outgoing links.

3. Use really obfusicating javascript OR iframes OR forms to do your external links. Suppose you have a blog post with 60 comments, put all the comments in an iframe and you mitigate the PR loss because the blog post just links once to the iframe, so if you have 20 links on your menu only 5% goes to the comments, instead of 75%. With javascript, Google can crawl basic javascript links, so you really need to mess them up. Hybrid form navigation is pretty safe.

4. Possibly use noindex or robots.txt, I’m insure if these will still work, but if they do, they’ll be good options. You can send external links through a redirect script blocked by a noindex, thus preserving the weight.

5. Box all links within a redirect warning page. Few sites do this, but it’ll work. Make a redirect page link as you would otherwise, but make it visible functional HTML, with your header, and menu, and footer, and everything else. Have this link say “You are about to leave this site for an external link. We cannot vouch for this content.” They provide the link again or use a meta redirect on a timer or javascript to forward. So suppose 75% of your link weight is going to your comments, but then the comment pages go to these gatekeeper pages, which have 20 internal links and just the 1 external link. You’ve now mitigated your PR loss by 95%*d where d is the damping factor. Not bad, and perfectly white hat. I’ve never actually heard anyone mention this type of thing for PR preservation before, the only sites I know that do it do it for legal liability reasons. Maybe I should patent the idea? In anycase you heard it here first.

6. Like 5 above, increase depth. If external links are located deep in your site, more clicks away from your homepage or other sources of weight, the less weight they will get. For instance, if you have a “links page” (something I’d never recommend) make it a “links directory” instead, thus increasing depth. This is actually also related to #1, you’re putting your most important content prominent, and putting the stuff you don’t want to get weight deep.

7. Simply do not allow certain types of external links. Removing them from blog comments for instance.

The most important thing to remember is that PageRank management is a game of percentages. You absolutely do not want to poison your site with excessive external or nofollow links on a PageRank heavy or otherwise important page. Keeping your homepage free of such influences will be a big deal and a big first step. Also, if there is a repeating link on every page of your site, making sure that isn’t hurting you is also important. The deep links from articles or whatnot are not as large of an issue (so long as they do not vastly out number the internal links on any given page, like what can happen with blog comments).

In the meantime, for right now, all of you who have put rel=nofollow on footer links to privacy policies and whatnot, take them off. They’re hurting you.

Also, if you were using rel=nofollow to control duplicate content, use rel=canonical instead. This new canonical definition is something software developers of forums and shopping carts desperately need to start adding to the default versions of their software.

13 Responses to “Nofollow & PageRank Manipulation”

  1. Bryan  Says:

    Nice analysis Chris. I’ve been changing a lot of pages today as I have typically nofollowed privacy, contact and some about pages on my sites.

    Your stuff is standing the test of time and you should be proud of that. Your work is the hidden gem of SEO and not enough people know about it anymore.

    I did find it interesting that Matt seemed to validate to some point that linking to good sites could play a factor with this comment, ” parts of our system encourage links to good sites.” when addressing Page Rank hoarding.

  2. Chris  Says:

    Thanks Bryan.

    I think, and I continue to think, Google only looks at outgoing links as part of their spam filtering.

    They may look for abnormal linking patterns and may identify spam sites that way. If you pass this filter adding authority external links won’t increase your ranking (it is redundant, if they make your site more useful, more people will link to your site because of it, and your rank will increase naturally).

    So it is pass fail, and I’ve never seen an official comment to dispute my theory. They’ve said outgoing links won’t help your PR or your generic ranking at times, but that they do look at them as part of their crawl. I gotta think that means they look at them as part deciding whether or not to apply a penalty, as opposed to how high to rank you. One might consider those to be the same thing, but they aren’t.

    In anycase it isn’t something I largely worry about. I in general control outgoing links on all my sites, on many you won’t even find a single one, and I rank well.

    Or, to put it another way, external links may be part of an overall “naturalness” question google uses to look for artificial activity or abnormalities, it wouldn’t be the only part of such a question though and this would be a gate, to get you in the door. Rather than used as part of the ranking algorithm. Once you are “in the door” so to speak I wouldn’t think an outgoing link is going to help you any great deal.

    I think this is a good analogy to the meta description tag. Google says they use the meta description tag, but they don’t use it for ranking as people assume they mean when they say that. Instead they use it to generate site abstracts for the SERPs in certain cases. You gotta remember when Google says they “use” something it doesn’t always mean that having that item will rank you higher.

    And of course, we know Google looks at outgoing links to detect blatant spam like link farms, circular linking schemes, reciprocal links, linking to bad neighborhoods, paid links, etc.

  3. Paul L.  Says:

    Excellent post! I’ve been removing a lot of the “nofollow” attributes from my internal links today and will be working on how I use internal and external links on my sites in the days to come.

    One thing I would like to see is an example of what you mean by “obfusicating javascript” versus “basical javascript links,” as I am definitely thinking of using javascript to “hide” links to low-priority internal pages such as “About,” “Contact Us,” “Privacy Policy,” etc.

  4. John  Says:

    Chris, FYI robots.txt blocked pages can and will accumulate pagerank – so those aren’t a good solution.

    Can you expand on this: Also, if there is a repeating link on every page of your site, making sure that isn’t hurting you is also important.

  5. Chris  Says:

    A sitewide link john. IF there is a link on every page of your site, because it is in your menu or your footer, and your site has 10,000 pages. You in effect have 10,000 external links right there. That is a big hole.

    Keep the external links off your menu, footer, and header.

  6. Roland  Says:

    Chris, you don’t post often but when you do it’s pure quality.

    Many thanks, you’ve solidified the half-formed conclusions I’ve had since hearing about the nofollow/PR changes.

  7. Chris  Says:

    Paul, Google can crawl basic javascript links, through event handlers or standard JS. To obfusicate it you can do a variety of things.

    1. Include the JS from an external file which is blocked with robots.txt OR blocked with referrer or IP checking. I’ve done this for security before, you check to see if the included JS is being included from your local page by looking at the http_referer variable. If not, you don’t serve it.

    2. Or muddle it in such a way that makes it hard to read, such as

    document.write(“<a hre”+”f”+”=”+”ht”+”tp”+”://www.exa”+”mple”);

    There are far more complex methods as well.
    The downside of this method is accessiblity, you lose it.

    Try this out:

    It encodes the JS (encrypts it in a way) to really make it hard to follow.

  8. Rimbow  Says:

    exelent your post How many search engines use in determining the quality / ranking of a web page, from the use of META Tags, the contents of the document, the emphasis on content and many other techniques or combination of techniques that may be used. Link popularity, a technology developed to improve the lack of other technology (Meta Keywords, Meta Description), which can dicurangi with a special page designed for search engines is called regular or doorway pages. With the algorithm ‘PageRank’ is, in each page will be inbound link (incoming link) and outbound links (links keuar) from each web page.

    PageRank, have the same basic concept of link popularity, but it does not only consider the “number of” inbound and outbound links. The approach used is a page akan diangap important if other pages have a link to the page. A page will also become increasingly important if other pages have a rank (PageRank) refers to the high page.

    With the approach used by the PageRank, the process occurs recursively where a ranking will be determined by the ranking of web pages that rangkingnya determined by the ranking of web pages that have links to the page. This process means a process that repeatedly (rekursif). In the virtual world, there are millions and even billions of web pages. Therefore, a web page ranking is determined from the link structure of the entire web page in the virtual world. A process which is very large and complex.

  9. Mike  Says:


    I heard that Google considers a sub-domain a different website. So on my main website I have a link (on every page) to the website blog hosted on a sub-domain with “nofollow”…

    On the other hand, I have a regular link in the blog that points to the main site (I want to pass the juice from the blog to the site)

    Is it a valid approach? Is there any pagerank lost with the new calculations?

    Thank you

  10. Massagestühle  Says:

    Hi… we had a change from pagerank 0 to 2 on our project on October 23rd. Can anybody explain why? There has been no general pagerank upodate that day.

  11. Chris  Says:

    Google is constantly upgrading PageRank, there does not need to be a well known global update for your PageRank to change.

  12. Marco Demaio  Says:

    Hello, just arrived here thanks to a link from

    1st of all let me thank you so much because for the 1st time I finally understood the technical implications in Matt Cutts’ article about “PageRank sculpting”. You are really an excellent writer! I went back and read Cutts’ article again and still without your post I would have only thought for the nth time: “Oh ok, PR sculpting is not that important anymore, so who minds?!” But your clever syllogism after a perusal of his article makes evident to me now the PR EVAPORATION due to nofollow links.

    There is just one thing that make me still suspicious about having all this stuff clear (or at least quite clear).

    You say: “Well, suppose you used rel=nofollow on a login page because search engine spiders don’t need to login to your site. What exists on that login page? Typically your header, your side menu, and your footer. Previously it was suggested to nofollow such links, to prevent unnecessary drain. This is no longer a good idea”.

    But Cutts says clearly on the “PageRank sculpting” article: “There may be a miniscule number of pages (such as links to a shopping cart or to a login page) that I might add nofollow on, just because those pages are different for every user and they aren’t that helpful to show up in search engines.”

    So what should we do? I would rather follow what yous ay, because it convinced me, but why do you think Matt says what I posted above?


  13. Chris  Says:

    Don’t use nofollow, even on login pages. Yes, the page isn’t useful to users, it won’t get ranked for any good terms anyways, but if you do use nofollow, you will still be throwing link weight away.

Leave a Response

(Email field must be filled in)

Top of page...