Monthly Archives: June 2009

A few weeks back, we used Officeworks’ as an example of what not to do when it came to launching a new website. Their clear lack of a migration plan led to a poor user experience (visitors who found an old listing in Google were shown a really poor 404 page) and unfortunately lost investment for the business (as thousands of previously optimised pages in Google were soon dumped from the index only to be replaced by inferior pages from the new website).

Of course, we’ve been keeping a close on Officeworks since they first launched their new website in late April 2009 to see how it all unfolds. From our hands on experience helping clients like Webjet, Carsguide and Truelocal we know that you cannot ignore the history of the old website. Doing so, especially for a large ecommerce business that already has strong existing placement in search engines can lead to hundreds of thousands of lost visitors, sales and millions of dollars in lost revenue over months and years if it is not corrected.

While we have noticed some improvements since our initial post (Officeworks is now using a 301 redirect to direct visitors to the new product pages), we are now seeing some other troubling issues, likely due to the implementation of the new SAP ecommerce system and lack of QA testing prior to launch. These issues, if left as is will likely cause unforeseen headaches for Officeworks in the search engines for some time to come.

So what did we find?

1. Duplicate content – multiple instances of the same pages in the index as a result of session IDs
2. Inferior content – automated pages created by the search engine crawlers

Duplicate Content

Shortly after launch we discovered that Google was indexing multiple versions of the home page due to the session ID used in the URL. A session ID is the most common method used by websites to collect data about a particular visitor as they use the website. Most websites only store a session ID using a cookie, so crawlers will ignore them. Officeworks however has decided to place the session ID in the URL when cookies are not accepted. Unfortunately, this is notoriously bad for search engines, often resulting in duplicate content issues.

For large sites, several spiders from the same search engine may visit at once. As a result each will be given their own session ID. While the content may be identical, the URLs are different each time, leading the crawlers to interpret the content as duplicate and thereby suppressing the content in the index.

In other words, Google is seeing the home page as a separate page each time it visits the website. This issue can be seen clearly in the following example:

Here we can see three versions of the home page that Google has crawled and indexed, each with a different session ID appended to the URL. Even though it’s clear that Officeworks is not up to any dirty tricks, Google will still suppress the value of the home page in this case. We also found examples of multiple listings for many other category and product pages in the index. As we know from the previous post, duplicate content will make it more difficult for Officeworks to achieve top placement in the search engines, especially when it comes to some of the more generic terms around office gear.

The quick fix is to only use cookies to track session IDs. First party cookies (the ‘safe’ cookie used for tracking session IDs) are a requirement for using most websites today so Officeworks customers will experience no loss in usability. In addition, Officeworks should add the canonical link to all of the pages to ensure the search engines apply the value from the duplicate pages to a single source.

Note: To view these pages in person type in site:www.officeworks.com.au into Google and click through to some of the deeper pages (i.e. go past page 500 to see some of the more interesting findings).

Inferior Content

A further handicap from the use of session IDs is that Google is now indexing a high proportion of inferior pages that have little or no resemblance to the actual category or product pages represented on the site.

By following the links (with the session IDs appended) throughout the website Google is able to effectively add products to the shopping cart or wish list in effect creating additional and irrelevant pages of content for the index. We can see this result more clearly in the following example:

Here we can see that Google’s crawler is able to add products to the Officeworks’ wish list thereby creating new pages. If you happen to click the third listing in the results shown (you need to turn cookies off in your browser) you’ll find the products below added to the wish list:

Example of Wishlist page in Google Index

While most people will never see these pages (unless they specifically search for My Wish List) they are still problematic, especially when it comes to achieving strong search rankings. Unfortunately, as these types of pages are created and indexed they began to compete with other more valuable content on the website. The end result is that Officeworks will end up with thousands of pages with little or no actual content in the search indexes that pulls down the rankings for higher quality pages.

Since the launch of the website, Officeworks’ pages indexed in Google have increased from approximately 7200 pages to over 9500. While this increase would be beneficial if each of the pages indexed provided unique content, in this case, the inferior content will only complicate matters for Officeworks.

Conclusion

While I’m certain there’s been plenty of learning within Officeworks (especially as we’re beginning to see some of these issues like session IDs cleaned up), there are a few key points for the rest of us to learn from to avoid the same pain.

1. A migration plan is essential when you’re redesigning a website, changing domain names or just reworking the information architecture. If you fail to plan, you plan to fail.
2. Quality assurance is critical. Some of these issues could have been spotted prior to launch had Officeworks ran specific Q A scripts to test whether the bots are able to effectively crawl the website. At the very least, you need to monitor what the search engines pick up after launch to determine the best course of action.
3. Do not trust the technology vendor to deliver a product that is search friendly. Do your homework and ensure the system does what you’re being sold.

Of course, there are no guarantees that you’ll have a smooth transition. But if you take the right steps, you can manage the big issues that cause havoc and react quickly to any surprises that pop up along the way. By doing so, you can avoid becoming the next casualty and lesson.