Explaining Google PageRank

Back in the mid-1990s the idea of using links to help rank web pages had already occurred to various people. In 1996, Robin Li designed the system called RankDex, in which links were used to help rank pages. The same year, Sergey Brin and Larry Page began working on BackRub, the system that two years later would be released commercially as Google. There was also Jon Kleinberg's hyperlink induced topic search at IBM and various others. Much of this early work was based on the concept of citation count or citation analysis which actually dates back to the 1950s.

The idea was that you could figure out how important academic papers are by analyzing the number and type of citations of those papers and other academic papers. In fact, today, Google Scholar uses citation analysis to rank search results. But it's specifically PageRank that is remembered and PageRank that occupies so much time and energy in the SEO field. Why? Because it's Google's link analysis algorithm and Google is of course, the giant in the search engine world.

So what is PageRank? Well, it's a patented algorithm named after Larry Page that Google uses to assign a numerical value to every web page it indexes. There are two things to understand.First, that it's complicated and we really don't need to understand exactly how it works. And secondly, that it is not exactly what Google uses, more than 14 years after the patent was issued. And they're not saying exactly what they do use now. They say they use PageRank, but it's undoubtedly changed over the years.

So what do we need to know about PageRank? That Google looks at links pointing to a webpage and uses those links to calculate a numerical value. It considers various factors beginning with how many links point to a page. The more links, the more popular the reference page is going to be. But it also considers the PageRank of the page on which the link sits. A link on a high PageRank page is worth more than a link on a low PageRank page.So you can consider each link is acting like a vote or a collection of votes.

The more links, the more votes. But some links, those from higher PageRank pages, carry more votes than others. It's rather like when shareholders vote. Some shareholders have more votes than others, because they hold more shares. Google considers other factors too.It looks at how many links appear on each page. A page's votes are shared among links on the page. A link from the page with only one or two other links would be worth more votes than a link from a page with a same PageRank, but 50 or a 100 links.

In effect, all the links on the page share the votes, so more links means that each link has a smaller share. Now remember, PageRank is passed from page to page, not merely site to site.Every page in a website has a PageRank, and that rank can vary widely throughout the site.Pages within your own site pass PageRank from one to the other, just as they pass rank topages on other sites, when you link out. We should also consider the idea of TrustRank, a concept from Stanford and Yahoo! but one that's almost certainly used by Google in combination with PageRank.

The idea is that links from trusted websites are more valuable than links from sites with no trust. Links from major newspaper sites and government research sites for instance, are likely to be more valuable than links from Joe's Blog. As the links flow away from the trusted sites, from site to site, the trusted rank pass gets weaker and weaker. PageRank is generally expressed as a number from 0 to 10, because that's how Google publically presents the number.

Behind the scenes though, the number is not, cannot be, a number from 0 to 10. PageRank is almost certainly a logarithmic scale with numbers reaching into the millions. The PageRank numbers provided by Google are ranges within the logarithmic scale. We don't know for sure,but if it's a logarithmic scale with the base of say, 6, then a page with a PageRank shown 0 could have a true PageRank between 0 and 6. If the PageRank is shown as 1, then the real PageRank would be between 6 and 36.

For 2, it would be between 36 and 216, and so on. Of course, this also means that a PageRank of say 4 is not twice that of 2. It could hundreds of times more. Which also means it's much harder to increase PageRank as you move up the scale. What then does Google do with PageRank? What do the other major search engines do with their equivalent systems which they certainly must have? There's a lot of confusion here. Many people tend to think that the page with a higher PageRank always ranks higher in the search results, but that simply wouldn't work.

Google and the other search engines are looking at a lot of different variables in order to rank pages, not just PageRank. I think of PageRank as a tie breaker. If you have two pages that Google thinks match a search query equally well, the one with the higher PageRank ranks highest. But it's not going to rank one page higher than another based on PageRank if the page isn't as good of a match with the search query. This is all a little complicated, but we can simplify it. What does all this tell us from an SEO standpoint? It tells us that we need links, and the more the better.

Ideally, we need links from high PageRank pages, better still, high PageRank pages with very few other links on them. Ideally, we also need links from trusted sites too or at least from sites that are close to trusted sites in a web of links. It also tells us that links within your site are important. If your site has few internal links, then you may end up with a decent PageRank on the homepage, but really poor rank on internal pages, making them less competitive in the search results.

I'd recommend that you ensure you have plenty of internal links throughout your site, soPageRank is passed through. Some people undertake what is known as PageRank sculpting, using the anchor tags rel = nofollow attribute for instance to stop PageRank from flowing to one area of your site, so instead it boosts others. Google recommends against this and even changed the algorithm to make it harder to do. When you know Follow Links, you're no longer giving the votes those links would've had, had they not been nofollowed to the other links on the page.

It's also important to understand what PageRank doesn't tell us. Unfortunately, many people in the business have not got this message. It doesn't tell us that only links from high PageRank sites have value. Links from low PageRank sites still carry some value in particular, if they're properly keyworded as we'll discuss in a later video. It also doesn't tell us that your site must have a high PageRank in order to rank well. That's an it depends situation. It depends on how competitive the keywords you are targeting are.

In many cases, well optimized sites with lots of nicely keyworded links pointing to them can rank well even with mediocre PageRank. Sure, in an ideal world, all your links will be well keyworded, coming from high PageRank pages with no other links on them. In the real world, you take what you can get.