Ars Informatica
November 14, 2018
Home
Health Care Informatics
Web-based MySQL/PHP Databasing
Programming
Web Development
Favourite Software
Hardware for the Frugal Fanatic
Graphic Design and Image Processing
Free Scripts and Software
About Us
Contact Us
Links
 
 

 Article Feed for this site

Google PageRank demystified: How To increase your page's PageRank - Practical Tips

March 29, 2007

Google PageRank is, at its simplest, Google's score for your web site - how important it considers it, based on the number of 'votes' other web sites cast for it, i.e. the number of links to your site. The PageRank or importance of pages linking to your site further influences your PageRank, as well as the importance of that link on that page, i.e., whether it is one of many, one of a few, or the only outbound link.

Your PageRank score in turn determines whether a searcher is likely to find your site on a Google search - the more important your site, the higher it will place in the search results, and the more hits your site will receive.

If you're looking for web traffic - and most of us do - you'll want to optimize your PageRank. And while PageRank may be Google-specific, other web engines almost certainly use similar algorithms, so optimizing for PageRank still makes sense.

This is Search Engine Optimization (also referred to by the abbreviation SEO) - or one good chunk of it. Some SEO techniques are intended to artificially inflate a site's PageRank, activities that can get you black-listed, and that aren't be covered here. This article explains PageRank and its workings, and gives concrete advice on how to logically implement your site - in a way that makes it the best possible value for your visitors, and optimizes its PageRank.

There are other articles on Google's PageRank, full of complex math and complicated modelling and eigenvector centrality stuff. If you're interested, I recommend the PageRank article on Wikipedia and The Google Pagerank Algorithm and How It Works by Ian Rogers.

This article still covers basic theory, and a few examples, but its main intent is practical. Real-world recommendations and worksheets round it out.

PageRank was invented by Lawrence Page (PageRank, get it?) and Sergey Brin, at Stanford University. Find the original paper at http://infolab.stanford.edu/~backrub/google.html. The basic formula:

basic PageRank equation

where PR is the PageRank, and PR(A) represents the PageRank of a certain page A, i.e. for any page you want to calculate this score for.

the other values:

    d is the so-called damping factor, typically set at 0.85
    n is the number of pages that link to page A
    T1 is the first page to link to page A, and Tn the last
    C is the count of outgoing links on any given page

The damping factor is an adjustment factor (typically set at 0.85) based on the knowledge that, on average, for any given page, someone browsing the web has only a 85% chance of continuing to click on to a next page. How it works is beyond the scope of this article. For practical purposes, however, it has several implications:

  1. The starting PageRank of any page, i.e. one without any incoming links, is 0.15 (i.e. 1-0.85). Every page has some intrinsic worth
  2. For any page linking to page A, the increase in page A's PageRank is 85% of the referring page's PageRank, divided by the number of links on that page

Regardless of what Google's damping factor is: every page - and every incoming link - has some intrinsic worth. You can't change your page's inherent PageRank, but you can get incoming links ...

Thus, the actual impact on your web page hinges on the sum of the values of all pages linking to yours, i.e.

equation: change in PageRank based on incoming links

This equation may look more complex; it isn't. Greek letter delta or Δ signifies the change in PageRank, and sigma or Σ indicates the bracketed part following it is to be summed for every page i from 1 to n. That is, the potentially modifiable part of your PageRank is equal to the sum, for all pages, of the PageRank of the incoming page, divided by the number of outbound links on that page.

For any given page linking to yours, your increase in PageRank value is (approximately)

PageRank equation: effect of one incoming link

Approximately, because, for the sake of simplicity, I've removed the damping factor. Additionally, it is likely that the actual PageRank is less than that calculated by the original formula, that it is a logarithm of the result - as discussed below.

There are obvious implications:

  1. The higher the PageRank of the incoming link, the better for your PageRank
  2. The fewer outbound links on the referring page, the better for your PageRank
  3. High PageRank effect is significantly diluted by many outbound links
  4. Low-ranking pages with numerous outgoing links are essentially useless

How useless depends on how little an increase in PageRank you consider worth pursuing. For example, a page with a PageRank of 1, with 20 outgoing links, one of which links to your page A, will contribute add 1 ÷ 20 or 0.05 to your page's PageRank - probably worth it. Twenty such links will increase your PageRank by 1, which may seem like little, but makes a significant difference in traffic. On the other hand, a page with a rank of 0.15 and 50 outbound links would increase your PageRank by 0.003. Worth pursuing?

Maybe.

PageRank ain't everything. One more observation:

  1. If you expect a link will bring you significant or quality traffic, it's worth pursuing, regardless of PageRank

Thus, the accompanying Web Site PageRank Assessment Worksheet, free of charge, in both Excel - where it auto-calculates the PageRank values, and helps decide whether a link is worth pursuing - and in PDF, where you'll have to do the math by hand.

You'll need to know the prospective referring page's PageRank. This is relatively easy - there are a number of utilities that will show you Google's last official PageRank for any given page. I recommend the PageRank utility included in - even if you don't use Firefox normally, it's worth having. You should be checking your pages in the most common browsers - Internet Explorer and Firefox, and maybe Safari.

Don't be surprised that most pages have a rank of zero - most pages have few quality inbound links. The reported score is a rounded value, and is updated at very variable intervals: from 18 to 122 days apart, since 2000 - though your actual PageRank, the one that Google uses to determine Search results, changes day-by-day. Still, the PageRank tool is a good starting point.

Next, check how many links that page has - including the ones it uses to navigate its parent web site, and image/ad links. This can easily run into the hundreds, though it may seem far fewer. If you really want to know, you could view the page's source code, copy this to a competent text editor (i.e. one that will count all occurences of a case-insensitive search string), and 'Find' all occurences of 'href', i.e. of all hyperlink tags.
 

Multiple-page sites, internal links, and the logarithmic down-ranking of PageRank

Every page has intrinsic value; it would seem logical that sites with many pages would increase each other's PageRank. They do. Still, you can hardly link all your pages to all your other pages: impractical, and something no visitor would appreciate. Additionally, you'd spread that page's "vote" over so many other pages that it would have little effect on any given page.

Most pages, however, have a menu structure: a top-ranked main page, and multiple second-tier menu pages, that link in turn to articles, blog pages, etc. These are your most important pages, these link your visitors to the content on your site. An attractive, well-designed menu system, and well-designed main pages are worth gold.

For a hypothetical menu system consisting of some 20 links to 20 main pages, with the menu present on all content pages, all with a base PageRank of 0.15, each content page would increase the PageRank of the main page by 0.85 x 0.15 ÷ 20 = 0.0064. Not much, perhaps, until the number of content pages starts to rise:

number of content pagesPageRank contributed by
menu to main pages
calculated PageRank
(0.15 plus contributions)
10
100
1000
10000
0.06375
0.6375
6.375
63.75
0.21375
0.7875
6.525
63.9

Impressive! and improbable. A PageRank of 64? Don't think so. Not even Wikipedia's English-language main page manages more than 9, as of this writing. And thus rampant speculation, on the internet, and in many a PhD thesis, that the change in PageRank is logarithmic - usually assumed to use a base of 6, that is, for every 6x increase in calculated value, actual PageRank goes up by 1.

Adjusting for this, we get the following PageRanks:

number of content pagesPageRank contributed by
menu to main pages
calculated
PageRank
logarithmic
PageRank
10
100
1000
10000
0.06375
0.6375
6.375
63.75
0.21375
0.7875
6.525
63.9
0.11
0.32
1.13
2.33

All in all - far more reasonable.

The adjusted PageRank formula:

logarithmic PageRank equation

(If you care - to get a PageRank of 0 to normalize to 0, add 1. So, 1 - 0.85 + 1 = 1.15. Also, I wish I could derive a simple formula for the effect of the logarithmic conversion on a single link - can't be done. But realize that the actual PageRank gained per incoming link is less than that calculated.)

Applying this equation to what we know about Wikipedia's English site: Main Page PageRank of 9. 1,708,229 articles or content pages, all linking back to this main page. A typical page with a PageRank of 5, 130 links (and many pages have hundreds), only 9 of which with the "nofollow" tag, and 2 back to the main page.

Applying the formula:

PR(Wikipedia Main) = log6(1.15 + 0.85 x (1708229 x 2 x 5/121))
                            = 6.5

Seems low, but doesn't take account of the hundreds of thousands of external pages that point to Wikipedia, as well as the effect of the foreign-language Wikipedias on the English-language site. In that context, the math seems reasonable.

Which means that routine-content pages, presuming they're 1 of 30 links on a main page (20 for the menu, 10 for content pages), or one of a great many more, work out to have a PR of log6(0.15 + 0.11/30 + 1) = 0.08. Essentially - zero. Zip.

Finally, based on the adjusted formula, to increase your PageRank from 0.15 to 0.16, i.e. by 0.01, you'll need to increase your base PageRank by 0.025. As PageRank get higher, this amount will increase accordingly. Keep this in mind when you set your worksheet thresholds.

Implications:

  1. The actual PageRank is always less than that calculated using the basic formula, and
  2. The higher your PageRank, the less impact increasing numbers of incoming links will have
  3. Don't direct incoming links to routine content, if not necessary
  4. Concentrate your PageRank: focus your menu links to your main pages, only
  5. Direct incoming links to your site's main page where you can
     

Next page in this article: Google PageRank demystified, Part Two: Targeting incoming links, PageRank dilution by outgoing links, and Reciprocal links and their effect on PageRank

Last page: Google PageRank demystified, Part Three: Links pages and local PageRank, rel="nofollow" tags, Modifications to Google's PageRank algorithms, and Recommendations

Copyright © 2018 Ars Informatica. All Rights Reserved.