Whenever I try to explain the concept of true Pagerank – not the fake number you see in the Google Toolbar – I find myself going into all sorts of metaphorical gymnastics. PageRank is like a tree… no, it’s like a fountain… no, wait, an electrical grid… or is it a squid…?
At long last, I’ve hit on a metaphor that works. It requires pipe, water, and some goldfish.
Stay with me…
Your web site is a pipe
Imagine your website is a pipe, carrying water from the bucket that is your home page to some happy goldfish that are at the bottom. PageRank is the water in the pipe.
Your home page concentrates most of the PageRank (water) because it typically gets the most links.
As long as enough water makes its way down the pipe, the fish are happy:
Links make holes in the pipe
Every link you place on a page is a hole in the pipe, diverting the water elsewhere. Those holes might be important – they aren’t necessarily bad. They might even be watering flowers or something. But if you’re the goldfish, you don’t care:
More holes means less water for you.
PageRank is a precious resource
The real lesson: PageRank is precious. Don’t waste it. You can steer PageRank around, just like you can pour water through pipes. Smart site architecture sends PageRank (authority) where you need it.
You want as much PageRank as possible flowing to your most important pages. The more links you have on every page of your site, the less of this resource you’ll have getting to those pages, and the less you can control it.
Links leak PageRank
If you’re a client, or a 10Things client, you’ve heard me say “you’re leaking pagerank” more than once. Now it should make a little sense. Links ‘leak’ Pagerank – they draw away some of the authority of a page, leaving less authority to ‘flow’ to other linked pages.
So link wisely, and consolidate links whenever possible. For example:
- Say you have a blog, and a monthly archive list on the right-hand side of every page that extends back to 2008. That’s almost 36 links on every page. Instead combine 2008 and 2009 into a single link for each year. Then list all posts for the relevant year on the target page. You just reduced 24 links to 2 links instead. That’s a lot of closed leaks.
So remember: Home page is a bucket. Like water, PageRank is a limited resource. You control the pipes. Be smart about it.
Wow, something I can understand AND take action on. Many thanks.
Thank you for the funny infographics and clear explanation. I still have the following question: can I also use nofollow for links to close leaks?
Thanks for helping me fix my seriously leaky blog!
@Inge definitely not. Nofollow doesn’t fix leaks – it causes the PageRank ‘water’ to evaporate, so you lose it altogether.
I thought that was the point of the nofollow. To control the flow of water/PageRank?
@LJ No, nofollow burns PageRank. It doesn’t control or sculpt it.
Thanks Ian. I don’t like the sound of burning page rank though. Is robots.txt any better at controlling the flow? Or does that burn it too? For Legal pages? And for nofollow? Do you only use that in cases where you REALLY don’t want to vouch for the site you are linking too? Because it seems like you might as well share it, instead of burn it, right?
This has motivated me to take a closer look because I think there are some leaks on my ship too. Great, another project!
Thank you providing an easy explanation on a topic that often hurts my brain to dissect, PageRank.
Interesting way to explain things 🙂
I espcially like the holes in the pipe – have for a long time searched for a way to explain the “leaks of PR” caused by outgoing links.
@Ian, Thank you for the feedback on nofollow issue
I follow most of this metaphor except for one part. Who/what are the goldfish and why do they need the water/pagerank?
Good analogy! Did people know that pagerank is named after Larry Page one of Google’s founders?
The tip on a blog’s archives was very useful, thanks!
That is one of the best explanations for Page Rank I’ve read Ian. Thank you!
Your analogy is flawed. PageRank is not a ‘resource’ to be hoarded. In fact, giving away links can increase your own PageRank.
I’ve tried to explain it at this post:
PageRank is simply the answer to the question: If you randomly surf the web, what fraction of the time will you spend at any given page? That’s the PageRank of that page.
Giving away links to related sites can increase your centrality to the network and, therefore, the percentage of time surfers visit your page.
@Bradd I’m not 100% clear on the connection between time on page and PageRank as defined by the original paper, but I hear you.
My example is super-simplified. However, the basic premise holds true – more links on a page will diminish that page’s PageRank as represented by the mathematical formula.
But you’re totally right. You still need to link to stuff – that’s how the whole thing works. But you need to do it wisely.
So am I better to have links on a 2nd page or a separate page of links… meaning, so long as they aren’t on the home page?
@LJ the best way to manage authority on your site is to create a good, faceted architecture. Anything artificial, like nofollow, could get you into trouble eventually.
I like your last statement, too: Why not share it, instead of burn it? Totally makes sense.
> @Bradd “I’m not 100% clear on the connection between time on page and PageRank as defined by the original paper, but I hear you.”
Time-on-page does not matter. Perhaps I should have said “the fraction of visits to page X is the PageRank of page X”. So, if there are 1000 pages on the web, and a random surfer visits page X 100 times, then the PageRank of page X is 0.100.
> “more links on a page will diminish that page’s PageRank as represented by the mathematical formula.”
I’m sorry, that’s simply not true. It’s a very common misconception, but more links from a page does not diminish that page’s PageRank. Again, I’m sorry if I’m not explaining clearly.
The short lesson is: Use links to make your page central to the network of pages on subject Y. Don’t be stingy trying to ‘preserve’ PageRank.
@Bradd I think you’re referring to PageRank’s iteration and damping as far as visits to page X, yes?
On your second item, that links from a page don’t diminish a page’s PageRank – I misspoke. More links from a page diminish the amount of PageRank that page can pass in each link. The page itself continues to have the same amount of PageRank. It just distributes it differently.
The rational surfer model affects this, but we don’t know exactly how, so I’ll leave it out of this.
Thanks for the insightful post, and the good ideas about limiting the number of links on your page. Do you think there is any merit to the idea that linking out to high quality sources (say, wikipedia, amazon, or other majorly authoritative sites) can improve your overall ranking?
@Kyle Yes, that can affect your ranking on Bing.
> @Bradd I think you’re referring to PageRank’s iteration and damping as far as visits to page X, yes?
No. Google uses an iterative method to calculate (actually, approximate) the PageRank of each page, but in the end, the PageRank of page X (according to the random-surfer model) is the fraction of all page visits by the surfer that land on page X.
This definition of PageRank is the same regardless of whether the calculation is done iteratively or not.
@Bradd K something isn’t coming through in the comment conversation we’re having.
Initial pagerank is approximated using a formula that includes, among other things, links into and out of a page. Then they include things like position of the link on the page, etc..
They use that to measure the likelihood of a visit to a page. They MAY track traffic to a page in terms of bounces back to Google. They likely track chatter. But actual visits to a page?
I’d love to talk to you more about this. E-mail me if you’d like to discuss. This sounds like a good blog post to me.
In general home-pages don’t link to other domains unless they have mutual understanding or some benefits(ads) the owner might be interested. Even in that case home-pages or any page of your site don’t necessarily link to another website’s internal page. Which means you link to the pages only to the sites which can bring actual benefits to your site(mostly traffic). And each time you say links can leak your PR, I thought you would have explained in that case about outgoing links rather links from the same domain.
Can someone explain me please, how is rel=’external nofollow’ different from rel=’nofollow’ and rel=’external’?
‘external nofollow’ is just an extra tag – the ‘external’ tells a spider that it’s a link to an external site. I’m not 100% clear on how consistently the search engines follow it, though, or the impact on SEO.
Fantastic post. I try not to get too technical with this stuff and really focus on the 20% that actually make s a difference. Well done with he simplification of this complex process…
Thanks a lot. I have wasted an inordinate amount of time looking for good info on SEO and related subjects, and finally I found your site. It is a breath of fresh air having some cater for a crowd other than a bunch of script kiddies
Loved your article;) My area of the web is strictly in graphics so concepts like these are hard to grasp and/or explain to my clients. What a cute and clever way of presenting the information in ways I can actually understand. Thanks!