iGB Affiliate 43 Feb/March 2014 | Page 31

TRAFFIC Avoiding Google’s Filters How to avoid link anchor text filters and other flightless birds, by forensic SEO specialist, Paul Reilly. Just the other day, my eldest daughter, India, was telling me she has to make a decision as to which subject she will choose for her GSCEs. She was considering history, something which I encouraged; while I was never a fan of history at school, I’ve come to realise in later life that history is such an important tool. In any evolving ecosystem, understanding the past enables us to accurately predict the future. If there was one crucially important skill in SEO it is this: you must instinctively and accurately anticipate Google’s next move. So as well as a brief delve into the recent historical algorithm archives, this article will provide you with insight into how I approach link strategy as well as providing you with fresh (at the time of printing) information from the MediaSkunkWorks data set which will help you in your link building efforts. How to avoid triggering an excessive anchor text filter In order to help you understand the method used, I’ll cover the first step. In order to make sense of large quantities of data, a simplified link profile is required. This requires us to group all page-specific anchor text data into the following groups: ●● Brand – e.g. www.examplecasinobrand. com, example casino brand ●● Exact Match – the exact keyword/term, e.g. ‘online casino’ ●● Phrase Match – keyword as part of a larger phrase e.g. ‘the best online casino’ ●● Other – none of the above (often call-toaction or non-transactional query e.g. ‘click here’, or ‘visit site’) Before we get into more of the analytical method, here’s a little piece of history. The first time I came across any solid evidence of an unnatural anchor text filter was around April/May 2010. This filter appeared (based on my own observations, which I must also caveat citing Heisenberg’s Uncertainty Principle) to punish specific pages, for specific keywords, where anchor text repetition appeared in abnormal proportions. This coincided with the release of an optimised crawling infrastructure, code name: Caffeine, and just to make the algorithm changes even trickier to isolate, Google also introduced something called the ‘May Day’ algorithm on May 1. Google messing with our minds We know this May Day algorithm wasn’t related to the filter I had been observing due to the timing. Rather than seeing the impact on or around May 1, it was clear that there had been two whole months of increased SERP volatility starting in early April 2010 and finishing late in May 2010. Within a large SERP dataset, I had observed numerous single keyword, page-specific ranking drops; these drops varied in severity and appeared to take into account the commercial impact of the positional change in order to limit the volatility of the results. If the observations were true, it was a significant change in policy for Google’s spam team. Previously, there had been fixed penalties such as 30 position, 60 position, and brand-based penalties. Previously, Google had a big problem when it came to punishing spammers. In highly monetisable SERPs such as ‘Online… Casino/Poker/Bingo’ as well as the usual tricky suspects, ‘Cheap Flights’, ‘Holidays’, ‘Car Insurance’ etc, everyone was cheating the system. Some agencies would even dominate entire SERPS, yet Google still wasn’t in any position to punish all the spammers. Hidden in the crowd of spammers Given that every major brand continued to buy links, punishing them all would be harmful to Google users. Not seeing the big brands on page one would lead to more users moving to alternative engines, like Bing. It was the presence of those big brands which indicate to the user that Google was returning the best results. We know this to be true based on another history lesson. Only a year earlier, when Google implemented the ‘Vince’ update, an artificial boost applied to big brands. According to Google, this ‘update’ only impacted a small number of queries and was never referred to by Google as an algorithm change. It turned out to be an introduction of a ‘white list’ to boost some big brands, who by the very fact they were playing by Google’s rules had been displaced by the aggressive brands and an increasing number of thin affiliates. Google had previously denied the use of such lists, but it was only when Eric Schmidt admitted under oath to the DoJ that white lists were in fact used, that it was widely accepted by SEOs. Previously, it had been a point of contention. Now, everything had changed… Google had finally solved the problem by punishing sites on a granular level (page-specific, keyword-specific), and now, the severity of the punishment appeared variable based on commercial impact. More recently, Google has referred to this type of penalty as a ‘granular penalty’. The first iteration of the granular penalty The filter appeared to trigger a granular penalty in cases where abnormal exact match anchor text had been used. I also made a second observation. It now appeared that keywords or phrases either in isolation or in broad-match groups – i.e. ‘casino online’, ‘online casino’ – could also iGB Affiliate FEBRUARY/MARCH 2014 31