Web Design is our .Biz: Google PageRank Prediction Exposed

What is PageRank?

“Devised by Google, it measures not only how many links point to a website, but the "quality" of the sites providing the links.” It is a ranking system on a scale of 0 to 10 (0 being worse, 10 being best) to determine how valuable or popular a website it.

What is PageRank Prediction?

PageRank is one of the holy grails of the search engine optimiser – they spend countless hours tweaking a website to increase their PageRank and their position in search engine listings. The PageRank of websites indexed by Google are generally updated every 3 months, so Search Engine Optimisers (SEOs) want to see if the tactics they have been trying have had any effect on their PageRank and turn to tools which try to predict their future PageRank.

One of the most popular tools to predict PageRank is by iWEBTOOL. This article explains why their tool does not work, and in fact, analyses the wrong data to come to its conclusion.

iWEBTOOL PageRank Predictor – how it works

There are three steps to how iWEBTOOL predicts your future “PageRank”

It sends a query to Google with the URL you specify and Google returns a result which looks something like:

Rank_1:1:6 Rank_1:1:6 Rank_1:1:6 Rank_1:1:2 Rank_1:1:4 Rank_1:1:4 Rank_1:1:4 Rank_1:1:4 Rank_1:1:3 Rank_1:1:3 Rank_1:1:3 Rank_1:1:3 Rank_1:1:3 Rank_1:1:3 Rank_1:1:5

It takes the last digit for each 3 digit combination and adds them up

It then works out the average of these numbers and returns the result as your “predicted PageRank”

The tool makes the assumption that the last digit is the PageRank of the top 10 pages who link to your website. It takes the further leap to think that Google works out it’s PageRank by taking these numbers and working out the average.

Where the PageRank Predictor goes wrong

I decided to look further into this PageRank prediction system and found that an XML file is generated to display the results shown above. I won’t go into full detail here, but in this XML file there is a node called <RK></RK> which contains the third number in the above results. People have assumed that RK stands for the websites PageRank.

In actual fact, after looking at Googles XML Reference, it is clear that RK has nothing to do with PageRank, but rather the relevance of a website with regards to the search term:

RK Definition: Provides a general rating of the relevance of the search result

So what is the PageRank Predictor actually doing?

It is using data which has nothing to do with PageRank to calculate your future PageRank. It is the equivalent of me wanting to find out how many apples will grow on my apple tree this year by going to count sheep in a field... Utterly pointless.

A bit more in depth analysis

For the more technically minded I have completed a little experiment to prove all of this.

We are going to analyse www.iwebtool.com using their own tool to prove that its PageRank prediction simply does not work.

If we have a look at the XML file generated by Google when we test www.iwebtool.com we will find the following results (NOTE: different data centres will return different results, so make sure you use the link above):

The XML file contains the following sites in this order

1. http://www.iwebtool.com/
2. http://www.iwebtool.com/backlink_checker
3. http://www.programmingtalk.com/archive/index.php/t-21828.html
4. http://www.programmingtalk.com/search.php?do=finduser&userid=141567
5. http://forums.digitalpoint.com/archive/index.php/ (shortened)
6. http://forums.digitalpoint.com/showthread.php?t=37333
7. http://www.webworkshop.net/seoforum/viewtopic.php?p=81569
8. http://www.searchguild.com/printer/fm2/22319/
9. http://www.statbrain.com/www.iwebtool.com/
10. http://directory.google.com/Top/Computers/Internet/Protocols/DNS/Web_Tools/

If we do a search on the same Google data centre for "www.iwebtool.com" (including the “” ) it displays the following links in the top 10

So, now we know that all these sites mention http://www.iwebtool.com somewhere in the page, but does their PageRank match the node in the XML file? Let's find out:

(format: link number. XML file value for <RK></RK> | Actual PageRank)

1. 6 | 4
2. 6 | 5
3. 2 | 2
4. 3 | 0
5. 5 | 0
6. 4 | 0
7. 4 | 0
8. 3 | 0
9. 3 | 1
10. 5 | 4

Apart from number 3. (which is most likely a coincidence) we can see that the <RK></RK> value does not match the actual PageRank. This clearly shows that by trying to predict the future PageRank by taking the average value of <RK></RK> will simply give you the average relevance of the top 10 sites who mention your domain name in relation to your website.

In Conclusion

If it was easy or even possible to predict a websites future PageRank, then the system would quickly collapse as SEO companies discover how it works and exploit it. We would see websites shoot up the rankings and the whole Google system would be a shambles. There is a reason Google’s algorithms are kept a secret. At this point, the only way to find out your websites future PageRank is to sit tight and wait for their next update!

Web Design is our .Biz

Google PageRank Prediction Exposed