At the end of February of this year, Google released their first “Panda” algorithm changes, which were intended to flag individual pages that were low quality. We applauded the effort by Google, since Panda was designed to help clean up plagiarizers, content farms filled with shallow repetitive content, and spam websites / blogs from search engine results.
Their strategy was to analyze webpages on a particular domain name, and look for “signals” that indicated the page was of low quality. This included the actual html markup of the page, advertisement numbers and placement, content analysis (depth, structure, readability, errors etc…) and a slew of other signals that could tell them whether a page was good or bad. They then looked at all the pages of the domain, and used the number of “flagged” low quality pages to give the overall domain a thumbs up or thumbs down. This meant that pages would no longer be treated as independently as they once were, and a few bad pages could pull an entire domain name down in rankings.
We never felt that SharedReviews should be lumped with low quality sites since we have always taken plagiarism and low quality content very seriously. From a shoot first ask questions later approach to plagiarism, all the way to suspending accounts where we felt the author just didn’t have the skills necessary to publish content that would be of value to readers, we have taken a “responsible” approach to our participation in the online content and publishing vertical. Unfortunately, our experience with Panda has been contrary, and we have seen dips and losses of traffic that correlated perfectly with each of the different “Panda Updates” that Google has released since February.
The losses have been so bad that we decided to go back to the drawing board, and see where we went wrong. The first things that jumped out to us was the actual HTML markup of the site, rendering speed, and the difficult to use navigation. We decided to do a complete site re-design, which we worked diligently on for many months, culminating in the public release of the new site redesign over a month and a half ago. We also identified what we felt were pages with low quality content, and blocked all search engines from accessing them until the author updated and improved the article (roughly about 15% of all content published on the site). After releasing the new site that is more usable, faster, more attractive, more consumable, and without any markup errors, I can confidently say that this does not appear to be why we were flagged. We even contacted Google directly and have had a confirmation that there were no manual spam actions taken against SharedReviews, but they provided no further feedback outside of a “canned” list of items we already addressed in the new design. Unfortunately, we have yet to see any recovery for the significant efforts we made to make our site better. From a reduction in bounce rate, to more time on site, to faster page speed, our readers seem to have liked the changes, but the “Panda” hasn’t.
Looking at the handful of sites that have seen confirmed recoveries from Panda, there were no real consistent items they all addressed that we missed, as far as we can tell anyways. Looking at the biggest winners and losers from each of the algorithm changes, we saw sites like blogger.com gaining huge increases in traffic (a Blogging platform known for spam and low quality content), while other good quality sites were being hit and experienced penalties for no explainable reason (Good quality unique content, good page markup, good navigation and speed etc…)
Blogging platforms, similar to us, have many different authors, with tons of different content published in tons of different verticals, but weren’t being hit negatively… why? Well, we came to one answer, and that answer was subdomains. You see, search engines treat subdomains effectively as separate websites. So if subdomain1.sharedreviews.com/article1 was flagged as “bad”, but subdomain2.sharedreviews.com/article2 was good, the bad article wouldn’t negatively impact the good article because it was considered a separate website.
We could continue to keep all articles on one domain, and try to identify bad content on an ongoing basis, but we felt that policing this on the scale we need was almost impossible since our first attempts just haven’t returned the results we needed. This is why, like popular blogging platforms, we will be moving all authors to their own subdomain. Starting in November (the actual date hasn’t been set yet), each author will have a 2 week window to modify their account nickname, which will be the new base for their account subdomain. For example, my account nickname is “Peter”, and if I do not change my nickname all of my public profile pages and content urls will be moved from http://sharedreviews.com/url to http://peter.sharedreviews.com/url. This will allow for some very cool new features such as the ability to tie in your own webmaster tools or analytics access for improved optimization, but most importantly it will allow our best authors to shine on their own without a negative impact from authors who have not made the same effort, or do not posses the same level of skill to attract and keep readers.
The new publishing game that we’re releasing is heavily dependent on readers and traffic, and we hope the addition of new custom sub-domains will provide our best authors the ability to receive the recognition and monetary rewards they deserve.