“Softening” Google Panda via Co-Citation
At PM Digital, we always have our ear to the ground, listening for the first indication of changes to search engines’ organic and paid listings. This is why we often notify our readers of potential or just-published updates to Google’s Panda algorithm; helping our readers turn what is often diffused forum buzz into actionable insights. Monday’s post by Barry Schwartz at Search Engine Roundtable piqued our interest because it highlighted what many are talking about as a “soft” or “silent” impending Panda update. It’s a bit too early to tell whether, in fact, an update is taking place (the MOZCASTs from Monday and Tuesday show normal days) but I believe this early chatter, and the idea of a softening of Panda, may tie to a YouTube post Cutts made on April 2 about site authority vs. popularity. Whether there is a Google update actually occurring or still pending, what could the April 2 comments by Cutts tell us about the potential focus of a “softer” Google Panda update?
Authority vs. Popularity
Let’s start by exploring the April 2, 2014 content on “How does Google separate popularity from authority?” within the Google Webmasters account on YouTube. In this video, Matt Cutts responds to the difference in popularity versus the principle of PageRank authority. Simply put,
“popularity is some sense is a measure of where people go, whereas PageRank is much more a measure of reputation, it’s much more reputation of where people link, and there is a disparity there or else porn sites would have the highest PageRank and government sites would be very, very low within our ranking system”
When sources like Search Engine Watch note that “Google is going to try to determine more between the site simply being popular and the site being a topic authority”, they may not be sufficiently expressing the fine conceptual point of the video. We’ve known for years that incoming links are a better ranking signal than traffic to a site – that authority is better than popularity. The long-standing method for determining this sort of topic specific authority is link anchor text. (If hundreds of people link to your site with text that says “leather handbags”, you’re likely an authority on the subject.) But anchor text has been a part of Google’s algorithm for some time and, we assert, is somewhat of an easily manipulated, blunt object. When we consider that recent Google updates have penalized sites with a high percentage of non-brand anchor text, I think we’re tugging at a thread that softens/mitigates anchor text as the lone signal of authority and relevance.
Co-Citation & Co-Occurrence
The idea of co-citation and co-occurrence has been around for some time but was summarized very well in a March 2013 article by Haris Bacic on Search Engine Journal. Co-citation refers to the concept that, when one site (Site “A”) links to two other sites (Sites “B” and “C”), information on or about sites B and C are associated with one another regardless of whether or not they link to one another.
Co-occurrence is a concept related to co-citation that theorizes a word physically located near another will impact the relevance of that proximal word. Meaning, if a sentence says “You can buy leather handbags for a great price at Brand X”, Brand X could be associated with leather handbags via co-occurrence.
Hold those two concepts in your mind and look back to March 2013, we see Matt Cutts talking about softening the impact of the Panda algorithm by finding additional signals to help refine the way they judge the quality of sites in the grey area of the algorithm. This is very similar to the language he used in March 2014’s SMX West presentation where he announced Google is actively working on the next generation of Panda updates that will soften the algorithm, with the aim of helping small businesses that may have been negatively impacted in the past. Viewed linearly, Cutts has moved Google from looking for additional signals to actively working with those signals, softening of their judgment on relevant mentions’ topical authority. When, in the April 4, 2014 video on Authority vs. Popularity, Cutts says “We actually have some algorithmic changes that try to figure out ‘hey this site is the better match for something like a medical query’” and then that “a site that actually has some evidence that it should rank for something related to medical queries… that’s something where we can improve the quality of the algorithms even more” we can read between the lines and theorize that he’s not talking about anchor text, he’s talking about something more refined. I propose this next level of refinement in topical authority is co-citation and co-occurrence.
I Thought Penguin Had to Do with Links, Not Panda?
Traditionally, when web marketers think about Google’s rules on linking, they think of the Penguin series of updates. This is a fair association, but Penguin’s function was mostly negative – a series of red flags that resulted in ranking demotions. Panda is properly thought of as Google’s quality algorithm, which is focused on not only penalizing low quality sites, but also recognizing and rewarding relevance within Google search results. Co-citation and co-occurrence lie in this soft middle ground between the two updates – yes, they have to do with links but they also have to do with judging the relevance of a site to a particular topic. The evolution of the Penguin algorithm is to bring link considerations beyond earlier, “hard” anchor text analysis and penalties into a larger, softer consideration of semantic relevance (co-occurrence) and link building without actual links (co-citation). If Google announces the retirement or consolidation of Panda vs. Penguin updates in the near future, it may confirm our theory that the search engine has entered a new “fine tuning” phase in developing their algorithm.