PageRank was and is one of the most misunderstood topics when it comes to search engines. Despite its surface complexity it actually is a remarkably simple concept.

When Google returns results pages, it has basic high level criteria that it wants to fulfill. The first is that the pages listed should be relevant to the query the searcher typed, the second is that they should be sorted in some way so that good pages tend to rise towards the top of the results. “Good” is one of those terms that we all know what it is but can’t define specifically, so you often find it substituted by other words. “Important” is the one Google uses.

This second sorting stage is where PageRank comes in. At it’s simplest level PageRank is a number that facilitates this sorting. As a page has a capability to be more about a specific topic than another page, we generally find the results listed are not in PageRank order but that there is a tendency for the higher PageRank pages to do better overall.

Search Engines did this before Google. Google merely redefined “Important”. Whereas previously the measure of importance was the number of pages linking to another page (or site), PageRank took this one stage further. PageRank adds that if a page is important then its links to other pages should imply a higher degree of importance in the page linked to, than if the page was less important.To put it another way, one link from an important page can be worth the same as many links from less important pages.

The reason this hadn’t been done before is most likely that search engines hit a catch 22 situation before they even started. If the worth of a link from Page A to Page B relies on Page A’s importance, you must first have an assessment of Page A’s importance. You cannot assess Page A’s importance with this method unless you know the importance of all the pages linking to Page A. To do that you would need to know the importance for all the pages linking to the pages linking to Page A. Which eventually means you would need to know the mportance of every page on the web, including Page B which was what we were trying to work out in the first place! The founders of Google took a different approach, rather than trying to calculate values for the importance of each page, they created a calculation that would make values slightly more accurate. Meaning you could attach any value of importance to a page, run this calculation and be slightly closer to the true result. Run this calculation again and you’d be slightly closer still. Run it enough times and you end up so close to the true result that you can essentially say this is a numerical value for the page’s importance.

Anyone that’s casually read about PageRank in the forums, or who’s downloaded the Google Toolbar can probably tell you that that final PageRank number is a value between 0 and 10. This is a misconception. PageRank values are more likely to be small numbers like 0.0567. When Google released the toolbar they chose to convert the actual PageRank values to a different scale. Just to confuse us further they chose to make that scale non-linear. Which means that from the values of 1 through to 9, although probably directly related to actual PageRank, there is very little we can learn from the toolbar. To further muddy the picture, if a page is not in Google’s index then Google’s toolbar will make a guess at what it should show. In this scenario, the real PageRank is zero but the toolbar may guess 7, the number that counts in the ranking process is the zero. There’s a distinction I like to make between Toolbar PageRank and Actual PageRank. The only place where the toolbar gives any reliable data is at the extremes of 0 and 10. As long as you are aware of the limitations, the toolbar does have some uses as a tool but I personally have found it increasingly less worthwhile.

Having said that, there are other ways we can learn a lot about PageRank. Observation and analysis of similar algorithms allows us to deduce some fundamental things that must be true of PageRank. PageRank relies on links and links alone. For that reason when we look at using PageRank in terms of optimizing pages for the search engines, we are essentially talking about creating optimum link structure. We are looking to achieve two objectives. Firstly, to get enough PageRank in to the site to work with and secondly to get that PageRank to the pages that can most benefit from it.

This nicely fits into the two categories we can divide links in to: External and Internal. External links are what is responsible for the site getting or losing PageRank. Internal links are what enables a webmaster to place that PageRank where it can best be used.When asking for links to a site, there are a few PageRank things that might well be worth checking. The first is that the page that is going to be linking to you is in Google’s index by performing a search on the url. It’s important here to remember that PageRank is page based. If they will be linking to you from their links page then that is the page you check and not their home page. The second thing you will want to do is show a preference for pages with less links. Do this regardless of the page’s current PageRank and you will form a good strategy that will benefit you in the long run. If those links are of quality sites then you also stand more chance of getting direct traffic.

When linking out to other sites, you are almost certainly going to decrease the total PageRank across all pages in your site. Meaning you will have less to work with. There are ranking benefits to linking out to good pages and there are obvious user/visitor benefits so you essentially have to make the decision “is this link worth the cost”. That’s really a personal decision that will depend on the level of competition in the area the web site is in, but invariably when linking out to good sites the answer will be yes.There are ways to minimise the PageRank loss when linking out, which is where we come on to internal links. Traditionally internal links have been the most overlooked area of PageRank. When a page says another page is important by linking, the quantitative value of that statement is based on the importance of the page doing the linking. If that same page instead said two other pages are important by linking to them both then they each of the two pages would get half that quantitative value. The quantitative value of importance that a page can give to others is divided by the number of links on the page. Say you have a links page with 2 links to external sites and no others. This effect means adding a link to an internal page on your site decreases the importance given to those 2 external sites. Or to put that in a way that’s more likely to make you smile - you’ve decreased, by just a tiny bit, the PageRank of those other sites. Where has that tiny bit gone? Back through that internal link to a page on your site.

Internal links can do more for you than just that though. If you modified the internal link structure of your page and waited a month or two, you would see the PageRank of the pages change. This works because PageRank is page based, and not simply site based. Each and every page, wherever it is, has a PageRank and takes part in the PageRank system. By using various link structures you can determine where the PageRank goes in your site. The specifics of those structures are beyond the space I have in this article, but in general a page which you give more links to in your site will do better. Using this method you can target PageRank to specific pages.

This is the most important thing you can do with PageRank. There is only effectively so much PageRank you can pull in to your site, and unless you utilise that efficiently you might as well not have it all. If a page is already ranking well for a keyphrase, does it need more PageRank? Obviously not so targeting PageRank on to a different page that is having a tougher time is only logical. Often, people worry too much about getting their home page to a particular PageRank and completely forget that that PageRank is only worth anything if properly used. As a general rule of thumb you will find that if a page is competing for a competitive keyphrase or if the page is competing for a number of keyphrases, giving that page a greater PageRank is a significant plus. If a page is competing for one keyphrase or a small number of uncompetitive keyphrases it is easier for that page to rank and your PageRank may be better used elsewhere.

These days PageRank has once again been brought to mass attention, because of the SearchKing v. Google case. For anybody not familiar with the case, the situation basically goes like this: the owner of SearchKing sets up a business “PR AdNetwork” to sell text ads with pricing based on PR. Not too much later SearchKing, PR AdNetwork and SearchKing’s clients experience strangely reduced PageRank’s. SearchKing sue Google. Regardless of the specifics of the case, it raises some very interesting questions about PageRank. Google’s defence is to argue that PageRank is their “Opinion”. I guess the number one thing we can learn from that is that “it’s better to have the mathematical version attached to your site than the opinion version”. But one of the whole reasons why this question is important is because it for the first time gives us information about the reliability of what we are given. Who cares who wins when questions like that are about to be answered?! When Google displays their PageRank on the toolbar (with the wrong scale) does that represent an actual mathematical statistic or the opinion of Bert who’s sitting at the console looking for sites that say something he doesn’t like? And how does that difference change user perceptions of a site? If something is presented in the same sentence as the words “uniquely democratic” then we might not expect it to be manipulated by a few people’s opinions. To be honest, I think it’s a smart legal defence, but it does bring into question exactly what PageRank is and means.