When Google rolled out Panda version 1.0 back in February this year, in effect the game had changed. For years Google had been stating that quality content was the way to rank highly in Google, although we all knew, and actually achieved, quality rankings of even poor quality pages, simply with the acquisition of links. Let’s be honest and say that Google has been flawed for years and we all have profited.


Have You Been Pandalized?


Google issues penalties for a number of reasons so how do webmasters know if they have been Pandalized? Well there is an easy way to establish if a drop in rankings is due to a Panda. Basically, Panda updates roll out each month. If a drop in traffic coincides with one or more of the Panda Iteration dates below, then you are a victim of Panda.


Google Panda Iteration Dates


  • Panda 1.0: February 24, 2011
  • Panda 2.0: April 11, 2011
  • Panda 2.1: May 9, 2011
  • Panda 2.2: June 18, 2011
  • Panda 2.3: July 22, 2011
  • Panda 2.4: August 12, 2011
  • Panda 2.5: September 28, 2011
  • Panda 2.5.1: October 5, 2011
  • Panda 2.5.2: October 13, 2011



What Is Panda?


Firstly Panda is NOT a live algorithm. Panda is actually a machine learning background job which runs firstly on what many are calling a training dataset, with results exported monthly, to the actual Google search results. We already know that Panda targets duplicate content, and indeed pages with a high percentage of ads, but in reality Panda does far more. Looking at sites that have been hit, it is clear to see that many Pandalized sites, did not contain duplicate content, a high degree of ads or even poor content. So why were these site targeted by Panda?


If we are honest, many affiliate sales pages may indeed be well written, but in reality they are of little use to readers. For example if we take typical affiliate Pokerstars review pages, many contain exactly the same info. If you read 100 such pages that how many could we actually class as quality pages?


The key of course is trying to understand how Panda classifies good and poor content. I for one have been saying for some time that bounce rate must be a key factor or at the very least a starting point. I suggest you should read the article by Peter van der Graaf entitled Panda in detail. One section of the article states:


“This means that while bounce-rate (in this case: visitors returning to search results quickly) isn’t used as a direct ranking factor, it is used to teach the Panda new tricks. Signals like bounce rate are fed as bamboo to the Panda background system with the instruction to find out what patterns can be derived from characteristics that form thin content, unnatural text and excessive on-page advertising. The system picks various combinations of attributes combined to get a high degree of certainty for someone’s spammy activities”


Also when digging a little deeper I found a research paper by the man himself responsible for Google Panda, Biswanath Panda. The document gives an insight to  Massively Parallel Learning of Tree Ensembles including examples. I found the section detailing an experiment for computational advertising very interesting.


We measure the performance of PLANET on the bounce rate prediction problem . A click on an sponsored search advertisement is called a bounce if the click is immediately followed by the user returning to the search engine.


Ads with high bounce rates are indicative of poor user experience and provide a strong signal of advertisement quality.


The training dataset (AdCorpus) for predicting bounce rates is derived from all clicks on search ads from the  data size for various numbers of machines.


Each record represents a click labelled with whether it was bounce. A wide variety of features are considered for each click. These include the search query for the click, advertiser chosen keyword, advertisement text, estimated click through rate of the ad clicked, a numeric similarity score between the ad and the landing page, and whether the advertiser keyword precisely matched the query.


So from the above we can deduce that an ad clicked with the user quickly returning to the search engine is classed as a poor quality ad. We can also deduce that an ad receiving more than expected clicks, with the user not returning to the search can be classed as a great ad. It is also interesting to note that there are other considerations taken into account which I have been testing.


How To Beat Panda?


Firstly if your site contains duplicate content, poor quality content or is heavily laden with ads, then you are at an extremely high risk of being hit by the dreaded Panda. Also if Panda is indeed mirroring the example test, published by Biswanath Panda, then if you have a high amount of pages with a high bounce rate, and by high bounce rate I’m referring to clicks from Google search pages that result in the user quickly returning to the search page. Then again you are a prime target.


It seems a no brainer to me that we all need to work harder to increase our click through rates from Google search pages and indeed work far much harder in engaging with our readers. Pages with high bounce rates should be remove or updated so that the actual bounce rate is reduced. With affiliate sales pages you simply need to go the extra mile so that more users to actual read them.


Many will read this article and state that they have sites with high bounce rates that have yet to been hit. How can I explain this fact? Well remember in real terms Panda is still in its infancy. With each update it learns more and more, with the effect of hitting more and more sites. Also of prime importance is the relevance of links in regards to Panda. Many have suggested that the quality of a site’s backlinks profile can indeed raise the threshold, for now at least, in regards to being hit by Panda. At the same time though it must be stated that more and more types of backlinks are being discounted by Google, which explains why the last Panda update, left many sites with reduced rankings of 10 to 20 places. Perhaps the last Panda update simply also contained another algo tweak that did indeed discount specific types of backlinks?


The trouble is that during the course of this year Google has changed so much! Many suggest that Google has also given more “power” to authority & brand type sites in the form of a Vince 2.0 update, which does indeed seem credible. Most webmaster for the last few months seem to be blinded by Panda so all algo tweaks are quickly classed as part of Panda which is simply not true.


Recovering From Panda


The good news for those that have been “battered” by Panda is that you can indeed recover. To date I have managed to get 2 sites back from the dead, with a third site well on its way, or at least I hope so.


The third site was hit by Panda 1.0 and was hit extremely hard losing most of its 100 daily uniques from Google as you can see below.



My tactics for recovering from Panda are as follows:


  • Remove any duplicate content without any redirects. Use online tools such as Copyscape to check ALL pages.
  • Report any scapper sites that may have copying your sites content. Use this form.
  • Remove thin content pages or beef them up.
  • Remove pages with a high SERP bounce rate or improve the SERP bounce rate.
  • Add more quality pages to your site.

Also of great importance is the fact that many have also stated that once a site has been Pandalized, simply removing low quality pages has not been enough to get the site ranking again. If you look at the analytics above, you can see the site above did recover although was quickly Pandalized again. I believe it is simply not enough, in most cases, to simply remove low quality content. You have to raise the overall quality of your site once you have been hit.


The Future of Panda


As I have already said, Panda is just in its infancy. The days of simply buying Google rankings with links, will I believe, within the next year, be gone forever. Google has been flawed for some time with links having an excessive amount of power in the Google algo. The truth is many sites with quality white hat links have already been hit by Panda. Although many affiliates do not want to believe it, the game has indeed changed!


Comments are closed.