It has been over six months since Google shattered the hopes and dreams of webmasters everywhere when they pushed out their biggest algorithm change for years. The Panda update saw thousands of victims going public with desperate pleas for help and advice, some sites reporting a complete loss of Google traffic, had opportunist search engine optimizers touting for trade on message boards web wide, and had reputable search engine optimizers admitting that they had no idea how to help their clients recover.

Three months after Panda 2.0, which was released on February 23rd, I searched high and low for examples of websites which had made a full recovery from the set back. Things looked gloomy at the time, my search was fruitless. There were a few medium to large sized businesses reporting a partial recovery, but these provided little by way of conclusive evidence and those partial gains could well have been attributed to on-page factors which were holding those websites back even pre-panda.

The sites which took the most pain were multi-user sites, otherwise referred to as ’content farms’, their ilk include well known brands such as Hubpages, Squidoo, and About.com. The reasons were soon made pretty clear, Matt Cutt’s had acknowledged publicly not long after Panda that even a small proportion of low quality content could bring down an entire domain name, examples of poor quality content may include duplicate content, spun content, really short content, and articles written by somebody who uses English as very much a secondary language. The initial reaction by almost all of the worst affected ‘content farms’ – I prefer to call them open publishing platforms, seeing as I hold portfolios on these sites myself for their once superior ranking abilities and would prefer not to tarnish myself with the slightly self-depreciating label of ‘content farmer’ – was to begin the process of deleting large quantities of lower quality content.

That step seemed pretty logical at the time, algorithms often use ratios, it is well known that Google themselves use ratios to judge various indicators, and I think most SEO guys would have expected a reduction in the proportion of obvious low quality content to reduce the severity of penalties in some sort of correlation. In hindsight, and in light of various new pieces of information and the inevitable speculation, in turns out that Google had done things very differently this time around. They had taken a hardline approach, a zero-tolerance approach if you like, to quality standards.

So where do you go from there? It was inevitable that somebody, somewhere, would come to be known as the man who cracked the Panda algorithm. It was a pleasant surprise, a welcome one in fact, when it became clear recently that the person who can lay claim to that honour is somebody that I have had the pleasure to meet briefly in San Francisco. Step forward the CEO of Hubpages Paul Edmondson who initially went public with his initial findings to the Wall Street Journal in mid-July.

Edmondson claimed to have stumbled across the possible Panda cure after noticing that Google had indexed some of his websites content under ww.hubpages.com rather than hubpages.com, and that the content indexed under the former was outperforming the correctly indexed content. After doing so he emailed Google (Isn’t that nice? When you reach the top 100 global sites you get a direct line to Google apparently! That’s what it takes to enter the online illuminati.) and asked whether his site should be split into sub domains. To cut a very long story short, they said yes, so he instructed his engineers to test run a test using a small selection of user accounts. Within days of that July announcement a large proportion of the site had been shifted to sub domains, with each user getting their own sub domain just like a blogger or tumblr blog.

I was initially highly sceptical, Quantcast figures showed a marginal increase, and some users reported an increase in traffic. Most of them reported a partial recovery rather than a full recovery. My first thoughts were that those sub domains would run through the next Panda update (which at the time was 2.3) and then be considered mini-content farms themselves. Traffic growth site-wide was relatively slow, and the change did not appear to have had the impact that Edmondson and his colleagues (I’ve met most of them too, got some t-shirts, nice lot) had hoped that it would have. To be frank, the whole thing was a bit of an anti-climax. This may be a good time to tell you that I am one of the major traffic guys at Hubpages, and therefore have a possible bias opinion yet also have a decent amount of insight. Alexa.com is telling everybody already, so what do I really have to lose:

My own traffic saw a decent increase, perhaps 30%, this still left me far short of my pre-panda traffic levels; in fact I was still 50% down. I didn’t expect this traffic to rise any further, I saw it as a dead cause. I was naturally delighted two days ago to see upward pointing arrows all over my statistics pages and in Google analytics, I’m not the only one either. The vast majority of users on the forum are reporting huge traffic gains, and Quantcast is showing these gains too. Here are the latest Quantcast stats, as of 1:31 GMT on 27th August 2011.….

That’s a fine upward sloping graph if I have ever seen one. At the bottom you will see that Quantcast is reporting 577k US unique for the 24 hour period preceeding. That is still a fair way off the record 874k, recorded just two days before Panda struck, but it is a lot higher than the miserable 264k that I spotted whilst monitoring progress before content was separated into sub domains. Who knows how high that graph will keep on travelling, but there is no denying that the recovery is in full swing. I’m not going to show you my own traffic graph, but it looks even better than that, in fact my own traffic has recovered entirely.

As a result of the decision to move each user account to their own sub domain it seems that content judged to be of a very poor standard is now unable to hold back better quality content. This in turn enables some user accounts to be penalized for Google, and others to be allowed to flourish. Many users are reporting poor traffic levels, whilst many others are reporting traffic which matches or even exceeds pre-Panda levels.

This switch also allows for further analysis of data, potentially enabling the Hubpages team to identify ranking factors which may be holding back user accounts; there are bound to be reasons why some accounts struggle and others don’t. The failure to employ basic SEO techniques will always be one of course, but putting those firmly aside it is also possible to gain a greater understanding of just how Panda changed the ball game. Edmondson himself recently published an article, Characteristics of Content That Drives Content, which summarises his early findings. After seeing the recovery for yourself, he looks to be somebody who is well worth listening to right now, I will be watching keenly as the story continues to unfold.