Fixing Un-Canonical URLs. Oh joy! Part 1

Jordan Sandford - July 25, 2008

In my previous post, Filenames, host names and canonicalization, oh my!, I talked about how the duplicate content issues affect your search engine rankings, and specifically how un-canonical URLs create this issue. I mentioned different types of URL canonicalization issues you are likely to deal with in your SEO work. Since that post, I created a new, more complete list of canonical URL issues, and will go into a bit more detail in this and future posts describing how the issues actually arise and how to fix these issues.

Let’s start with a detailed guide on how to pronounce canonical: ca – NON – uh – cul. Also, canonicalization is pronounced: ca – non – uh – cul – i – ZAY – shun. That was easy!

Without further ado, here is the new list of areas of canonical URL issues:

1. Protocols (http and https)
2. Domain and subdomain names (sometimes referred to as host names)
3. URL paths
4. File names
5. Case sensitivity (when myPage.html is handled differently than MYPage.HTML)
6. Query strings
7. Combinations of any or all of the above

Let’s break these down using an example URL:

https://example.org/blog/Colors/default.asp?action=go&sessid=6468439#section3

1. https is the protocol
As far as search engines go, there is only http and https.

2. org is the top level domain

3. example.org is the domain name
No subdomain is used. (Technically, www is a subdomain just like any other subdomain.)

4. /blog/Colors/default.asp is the URL path (everything after the top level domain, including the file name, and before the query string, if any)

5. default.asp is the file name
Many times, the URL path may not contain the file name at all and will just end with the forward slash (e.g. /blog/Colors/).

6. ?action=go&sessid=6468439 is the query string (everything after the question mark and before the # sign, if any)
The query string is made up of parameters and their values. “action” is parameter and “go” would be its value.

7. #section3 is the URL fragment (also known as a named anchor or bookmark)
A URL fragment tells the web browser where inside a specific web page to go to or to scroll to. It does not tell the web browser (nor the search engine) which specific page to go to, so therefore does not contribute to any URL canonicalization issues.

To illustrate each of these, here is perhaps the worse case scenario as it relates to canonical URLs (I will continue to add to this scenario from post to post in this series):
You have a site that has forms visitors fill out. You have an SSL certificate and people can go to (https://www.example.com) to see your site. Now, you want to give your visitors peace of mind by letting them know that your site is secured (by making sure that browsers see the lock icon that indicates a secure site). When your https version of your site is setup, the web server may pull files from the non-secure section, but send them back to the browser over a secure connection. (This ‘usability feature’ provides the convenience of only having to maintain one set of files instead of two sets.) Now, when visitors go to https://www.example.com/blog, they’ll see the same content as http://www.example.com/blog. This is your first duplicate content issue.

There are different ways to handle protocols when dealing with canonical URLs. You can block all access to the secure version of your site using a robots.txt file. However, if your web server or web hosting account is setup to serve your secure site from the same files as the non-secure files, search engines will see the regular robots.txt file when going to https://www.example.com/robots.txt. The most flexible way to circumvent this is to create a version of the robots.txt file that’s used only for https and name it appropriately. Then use rewrite rules to internally redirect all requests of https://www.example.com/robots.txt to robots-ssl.txt. The search engines will still think it’s looking at https://www.example.com/robots.txt, and the industry standard. If your site is running on an Apache web server, see apache.org’s URL Rewriting Guides and yourhtmlsource.com’s URL Rewriting Guide.

Another way to handle this will only work if all your files on your web server can run server-side scripts. Usually, this is the case with .asp, aspx, .cfm or .php files. If you have .htm or .html files, you may be able to ask your web hosting company to allow server-side scripts to be ran in these files. Once all files on your web server can run server-side scripts, make a script that checks if the file was accessed via https, and if so, add a robots meta tag that disallows the page from being indexed. This script needs to go in every one of your files, either by ‘including’ them or pasting the script in at or near the beginning of the page. The file-include method can drastically reduce the time it takes to administer this script since you only have to make changes to the script in one place; all files that are including the file that has changed will be essentially updated automatically.

My next post will discuss canonical domain names, including subdomains, and URL paths. In following posts in this series, I will provide methods that maintain the usability features of the web (for both the webmaster and the visitor) yet prevent the duplicate content/canonical URL issues. Stay tuned!