Recently, I have been experimenting with canonical urls and how it relates to SEO. The results of my initial experiments are interesting. SEO experts have been told that using canonical urls are good practice by Google to make search engine results more relevent, and also that they help the site itself get the relevent page higher in the results.

As an experiment, I decided to test to see if adding canonical urls to a site really makes a difference:

Introduction:

Canonical urls are claimed by google to help search engines index the correct versions of pages, and to prevent the duplicate content problems found by having the same content on different urls of a site. However there are 2 different schools of thought as to whether this increases or decreases traffic to your site. On one hand, some argue that canonical urls help the page ranking for individual (stem) pages, and therefore increase their prominance in the search results. Others argue that removing similar pages from the index, reduces the number of in-site links, which reduces the rankings of the site as a whole. This experiment attempts to see which school of thought  is more likely to be correct.

Method:

The site chosen was www.onlineglassesstore.com - this site sells fashion glasses online and is suitable for this experiment because there are different paths that you can travel to get to the product details page.

The first step involved getting the duplicate pages indexed. This was not too difficult, and if you google the following: http://www.google.co.uk/search?q=site:www.onlineglassesstore.com+772&hl=en&filter=0 you will see that there are 3 links that point to the same product (as of writing, by the time you read this, duplicate pages may no longer be indexed).

Then the ranking of the homepage was measured for the phrase 'glasses online' (link included for SEO purposes). The site was at position 8 in google for this phrase.

Following this step, a canonical url was added to the site, so that all 3 versions of the page: http://www.onlineglassesstore.com/fashion/retro/772 , http://www.onlineglassesstore.com/designer-glasses/Four-Eyes/772 , and http://www.onlineglassesstore.com/prescription/lightweight-glasses/772 show the same canonical url, which was set as : [category]/[brand]/[product id] .

One week was allowed to pass, and the ranking of the site was re-analysed

Results:

1 week after the canonical url was included, the home page of the site slipped from position 8 to position 14 in google.co.uk for the phrase 'glasses online'. It was no possible to measure the individual pages rankings, since the site comes up top in google for 'lightweight glasses' already, but it is postulated that the pages might have seen a small increase in rankings for long tail keywords.

Conclusion:

It appears from this small test that the addition of the canonical url actually hindered the ranking of the homepage of the site. This had a great impact on the level of traffic, some 40% reduction in traffic and similar affect on online sales.

The theory as to why this may be is as follows: if there are 3 interior pages to a site that each link to the homepage, then the homepage has inbound 3LJ (link juices). If the 3 pages were amalgamated into a single page, then the inbound link juice for the single canonical page is up to 3 times the inbound LJ of the non canonical urls. Therefore the individual interior page has better rankings themselves, but the side affect is that the fewer pages linking back to the homepage makes the main page have lower page rank because there are less total link juice flowing in. Why less total link juice? well, it is likely that the canonical url does not have exactly 3 times the link juice as the 3 non canonical pages, it is likely a leakage factor is applied when transferring link juice from the duplicated pages to the main one.

Although seeming to support the canonical fallacy, it could be that other SEO factors have come into play which affects the rankings. Things like inbound links from external sites, domain age factors and content freshness are all part of the google ranking algorithm that aply even when no one makes any changes to the site. These are known a passive SEO factors.

We have now removed the canonical URL and will do another ranking analysis this time next week.

Watch this space for the results.