Why is this store getting hurt in SERPs when they removed duplicate content?

[merch_zzounds]

I work with an e-commerce client who got hit hard by Panda. They are very cautious, and want small-scale tests to prove each hypothesis before committing to larger changes.

Recently, we reworked content on 30 product detail pages. Before, these product pages featured some original content mixed with some manufacturer content. The change we made was to remove the manufacturer content completely from the product page, leaving about 300 words of high-quality, original content--all of which was written by subject matter experts.

I assumed that Google viewed this manufacturer text as duplicate content. However, when these 30 modified pages were compared to the control, they performed significantly worse.

Question 1: Does any have any idea why these pages would perform worse than the control?
Question 2: Do you have any tips for convincing this client to try another test or get the buy-in to make the larger changes that--in theory--need to happen?

FWIW, this client has about 10,000 product detail pages--the vast majority of which contain just manufacturer content.

I appreciate your thoughts.

[RyanKent]

I would be curious to look at each form of the pages using the LDA tool.

The main reason to use the tool is to understand if you are significantly reducing the page's relevancy to your target keyword(s).

[merch_zzounds]

Hi Ryan,

Thank you for your thoughtful reply.

We didn't make any changes to any of the titles or any overall site changes. We measured the changes using internal tracking tools to get the daily traffic by search engine to each product detail page. We also used Google Webmaster Tools to estimate SERP positional data.

The updated pages do have about half as many total words as the control, so that could definitely explain it.

For followup testing, I'm thinking of trying:

• test versus control, same word count, test is 100% original, control is 0% original
• test versus control, same word count, test is 50% original, control is 0% original
• test versus control, same word count, test is 25% original, control is 0% original

[RyanKent]

Hi Sarah.

I can't blame the client for wanting small tests. It's actually a prudent strategy.

I think having some manufacturer content mixed in with other quality content is fine. Normally 300 words would be a bit thin but for product pages that is not bad at all.

What changes to the page were made other then the content? Did the title change at all?

How exactly are you measuring the performance of these pages. Please provide as much detail as possible.

Is there any other site wide changes being made at the same time?

Does any have any idea why these pages would perform worse than the control?

Assuming there are no other factors involved, which is a big assumption, the only logical response I can offer is the revised pages are viewed by Google as weaker. By removing the manufacturer's content the pages are thinner and less relevant to the keywords involved.

Do you have any tips for convincing this client to try another test or get the buy-in to make the larger changes that--in theory--need to happen?

How to restore a client's confidence after making an error?

Assure the client you realize they are highly dependent upon their rankings for sales. You are proceeding under your experiences and knowledge, which has been successful in the past. Panda is a new thing and Google is still making rolling out new changes. The industry is still learning and adjusting to these changes. You are consulting with others regarding the unexpected results and you are committed to ensuring this project is properly completed.