This article shows how altering a website's navigation tree affects the Page Rank of all pages on the site. Page Rank is described by Google as follows: PageRank relies on the uniquely democratic nature of the web by using its vast link structure as an indicator of an individual page's value. In essence, Google interprets a link from page A to page B as a vote, by page A, for page B. But, Google looks at more than the sheer volume of votes, or links a page receives; it also analyzes the page that casts the vote. Votes cast by pages that are themselves 'important' weigh more heavily and help to make other pages 'important' Source http://www.google.com/technology/ In other words, the Page Rank Algorithm it is a circular calculation. If two pages start with the same page rank and are only linked to each other the effects have to cancel each other out. However if the pages are not equal, interesting things begin to happen. Because scores affect scores which in turn can affect the scores of the original pages a feedback loop develops. Therefore where you add your new content can have unforeseen consequences to the Page Rank of other pages on your site. We can use this knowledge to manipulate the boost in page rank produced by the addition of new pages. We can decide in advance if our new pages should boost the homepage or any other page or collection of pages. Let us imagine a small idealised website with 10 pages (imaginatively named a.html, b.html, c.html, d.html,e.html,f.html,g.html,h.html, I.html and j.html). Here is a sitemap for such a site. graphics embeded Here is the current ranking of the pages on this site. Page Rank A.html 1.8896 B.html 1.8896 C.html 1.8896 D.html 1.4724 E.html 0.6100 F.html 0.6100 G.html 0.6100 H.html 0.6100 I.html 0.2093 J.html 0.2093 This ranking was derived by applying 20 iterations of Google founders Larry Page and Sergey Brin’s page rank algorithm to the pages. Now suppose we add another eight pages to the site. we decide on three options for ways in which they might fit into the site. Option 1 We could add them all as extra pages at the same level as e.html to h.html graphics embeded Option 2 We could add them as a completely new level all below e.html like this graphics embeded Option 3 Or we could spread them as a number of sub pages like this. graphics embeded By calculating the estimated page rank for all of these options we can choose which will serve our SEO purposes best. Again by applying 20 iterations of the basic page rank algorithm to the pages we get the following results. Results table Page Rank Option1 Option 2 Option 3 A.html 1.8896 2.0500 2.7718 2.5649 B.html 1.8896 2.0500 2.7718 2.5649 C.html 1.8896 2.0500 2.7718 2.5649 D.html 1.4724 2.0500 2.7718 2.5649 E.html 0.6100 0.7142 0.9865 0.9085 F.html 0.6100 0.7142 0.9865 0.6910 G.html 0.6100 0.7142 0.9865 0.6910 H.html 0.6100 0.7142 0.9865 0.6910 I.html 0.2093 0.6145 0.7092 0.7163 J.html 0.2093 0.6145 0.7092 0.7163 k.html n/a 0.7142 0.2777 0.4157 l.html n/a 0.7142 0.2777 0.4157 m.html n/a 0.7142 0.2777 0.4157 n.html n/a 0.7142 0.2777 0.4157 o.html n/a 0.7142 0.2777 0.4157 p.html n/a 0.7142 0.2777 0.4157 q.html n/a 0.7142 0.2777 0.4157 r.html n/a 0.7142 0.2777 0.4157 Results narrative Option 1 had a weaker effect on homepage and top level category pages (a.html, b.html, c.html, d.html) but the newly added pages were ranked well. In Option 2 the boost to the homepage and top level category pages (b.html, c.html,and d.html) was most marked. However with this option the new pages themselves scored poorly. Option 3 is a compromise. Apart from the one page directly above the new pages (e.html), which does quite well out of the arrangement, all the other pages get a moderate boost. Conclusion With this knowledge we can decide where to place new pages depending on our needs. For example, if a boost to the homepage is our primary seo requirement we should add new pages as a completely new level all together and crosslinked with each other. However if the new pages need the traffic we should add them according to the map described by Option 1. Of course page rank is not the only factor in search results placement and page rank itself is strongly influenced by external links as much as internal links. Nonetheless, when looking for that extra edge that is needed to achieve an SEO goal, this information should prove to be of real benefit.
Great to see some posts with real numbers - more objective stuff like this and less subjective stuff is great to see. A few questions though: when the extra 8 pages are added the linking structure between the existing pages also changes (e.g. new links between B, C and D are added, also I to J, F to H). Since the idea is to look at how the new pages change things, this seems a bit odd. Would be interesting to redo the number where the only change is the addition of the new pages/links. And the starting point seems a bit unusual in that the "home page" A only has links to it from two other pages. Wouldn't it be more common to have links to A from all of the first 10 pages? One final thought - it's interesting to see that B and D have quite different PR in the initial setup. They both have a single link from home (A), and the only differences come from the number of sub-pages they have. This suggest simply adding more sub-pages (that also link back) means you get a gain in PR. I need to look at the formula to try and figure this out, but I was surprised to see this. Can you really gain PR this way? Howard
Thanks for the feedback Howard No changes were made to the top structure when the extra pages were added. The digram I posted on the website was incorrect and has been amended. Thanks Chris
Thanks for the update. There are still a few upper links in the diagrams that change when the new pages are added, but let me see if I understand the basic setup and the three options. Not trying to be a pain, just interested to see exactly what you tested as it's a nice bit of work. basic setup: as in your top diagram, but I assume there are also "sideways" links E-F-G-H and I-J in the initial setup (these are all there on the option 1/2/3 diagrams). Option 1: extend the sideways link E-F-G-H to join to K-L-M-N. The extra "curvy" links between E/F/G/H I'm not sure about - should these join to O/P/Q/R? The other change is extra links all from E to K/L/M/N. Option 2: This is fairly clear: extra links from E to all new pages (K-R) as well as internal links between the new pages. Option 3: Pretty clear from the diagram, E, F, G and H each get links to two of the new pages. With option 3, E, F, G and H look identical in terms of links: they all link to B, and two of the new pages. I'd expect PR to come out the same for E-H (as in the initial setup) so I'm surprised to see E with more PR than the others. What causes this boost for E? Howard