eCommerce Search Engine Optimisation: Duplication

Established brands typically operate online with a key asset balanced against an often crippling weakness: domain (aka backlink) strength balanced against technological inertia.

The inertia stems from often operating with an (expensive) legacy CMS or eCommerce system which is not a flexible as many of today's free CMS systems. Typically there's also a team involved in maintaining the performance of the CMS, especially when the brand has high traffic levels which push the system to its limits.

With technical resource tied up in maintaining the system, solving SEO issues is often shunted to the bottom of the priority queue unless a strong business case is given to explain why the work is valuable.

Just stating that the work is beneficial for SEO and improved rankings is insufficient in such cases, so how can we go about building a compelling business case for eCommerce SEO?

Lets take an example site: Wickes.co.uk.

Digging into its index profile highlights a number of issues. For example, although a Google site operator suggests that it has 116,000 pages indexed...

...we can see that only 998 are listed in the main index, with the rest remaining in the supplemental index.

See: http://www.google.co.uk/search?q=site:wickes.co.uk&num=100

We can actually see a number of causes of this duplication using Google's search operator chained with intitle and inurl operators:

• Isolating indexed pages with 'MDF Sheets' in the title tag and adding &filter=0 to the search parameter reveals 48 results, of which all but two should be retained: http://www.google.co.uk/search?q=site:wickes.co.uk+intitle:%22MDF+Sheets%22&num=100&filter=0
• Investigating closer we see that using different sort options within reviews appends different 'sortOrder' parameters to largely identical pages.
• The use of Flash popups creates 22 URLs containing no content per product: http://www.google.co.uk/search?q=site:wickes.co.uk+intitle:%22MDF+Sheets%22+inurl:%22flashpopup%22&num=100&filter=0
• Largely identical product descriptions for marginally different products (different sizes of MDF sheet, or special offer discounts, or if a product is in-store pickup or not) results in 17 different URLs (in this instance) for the same product: http://www.google.co.uk/search?q=site:wickes.co.uk+intitle:%22MDF+-+MDF+Sheets+-+Sheet+Materials+-Building+Materials+-+Wickes%22+-inurl:%22flashpopup%22&num=100&filter=0
• We could dig in deeper for more fundamental problems, but lets return to the issue of building the business case to sort out these critical SEO issues.

We have to draw the line between page duplication being harmful, impacting on search rankings, and organic traffic's contribution to eCommerce sales. Achieving that would allow us to put a fairly accurate figure on the £ impact of failing to implement the solutions we propose to the issues above.

Can we build out that case study?

Well, we can establish that the &filter=0 parameter reveals content that is considered 'supplemental' by Google, given Matt Cutts has cited it several times. We can also cite Google's statement that they have only removed the label for supplemental results, rather than the supplemental index itself.

We can also prove that Google only uses supplemental results for competitive search terms (such as in the form “buy [product]”) when there are no remaining listings from its 'main index'.

So we've established that duplicate content held in the supplemental index shouldn't rank strongly. We can also pull off the natural search traffic driven to those supplemental URLs to verify the lack of incoming traffic.

We can then pull the best ranking URLs for a sample set of terms - we're using the MDF product here, but the process applies to any product - from Google’s Webmaster Tools and map those positions against SERP Clickthough data which can be used to estimate the clickthough rate to the top fifty rankings for a search term.

We then either match that to actual traffic and reverse engineer the likely traffic passing through the number one position, or, if we are ranked in the top ten, we can take the 'impressions' metric from the Webmaster Tools, subtract the estimated PPC clickthough traffic and apply the clickthough ratio for the remaining organic traffic to determine the size of the prize.

We then look at the average conversion rate of our product pages, and average product value for our sample product search terms, then calculate the available additional revenue improving our rankings would generate.
At this stage, the weight of your authority as an SEO comes into play when estimating the impact of the improved change to your ranking. Obviously, you can only estimate improvement to the marketplace as it stands. If your competition suddenly ups its game, either by dramatically increasing its linkbuilding work or resolving technical issues of their own, then the impact of your changes will be lessened.

That said, in this case we're looking at issues which create something in the order of 48 * 22 = 1,056 versions of the entire Wickes product range - effectively the whole site - and you can be sure of dramatic SEO improvements when resolving it.