As the Web grows and Internet user behaviors change, Google’s understanding of what is important must also change as well. The search engine updates its algorithm almost daily and works to return the most relevant results to its users for a given query. Additionally, because of spammy content infiltrating search results, what returns as a relevant result today may not do so tomorrow.
While certain algorithm updates receive more fanfare than others (e.g., the recent “Farmer / Panda” Update), Google is said to manipulate its search results more than ten times per week. These updates reflect what users expect to see in search results, and also how information circulates the Web.
The Impact of Algorithmic Changes
Recent algorithmic changes have focused on content farms, as well as social signals, to determine relevancy. The most recent update targeted content farms, such as Huffington Post and Demand Media sites (eHow, LIVESTRONG, Cracked, and others), and worked to reduce the rank of sites that benefited from contrived or duplicated user-generated content.
Regarding social signals, Google has been visibly working to integrate social into search – not just as a traditional ranking factor, but also by displaying tweets and tweeted links within its universal searches. Google Social Search, as well as the newly-announced +1 feature, represent attempts from the Internet giant to integrate social sharing in a way that will improve relevancy and encourage a more connected Web.
Reacting to an Algorithm Change
SEO tactics and best practices can grow stale quickly. With all the changes that are rapidly occurring, questions and concerns arise regarding what works and what doesn’t, and what constitutes ‘good SEO’ versus a bad practice or ‘black hat.’ This can put your SEO strategy on thin ice, and negatively impact not only your rankings, but your search traffic, as well. The obvious example of fishy SEO implementation was by J.C. Penney, which was publicly lambasted for deploying seemingly dirty tactics to raise ranks.
What’s Changed from ‘Traditional SEO’
In previous years, SEO was a more approachable task, with a much less crowded marketplace. To rank well in Google, it was rather simple: make your website’s structure easy-to-read, create good content, and shop it around to blogs and relevant Web directories. High rankings resulted, as well as boosts in traffic. While this basic strategy still holds true, application has changed significantly.
Changes in Content Marketing
Keyword density, defined as the percentage of times a keyword or phrase appears on a page compared to the total word count, was the only metric that mattered for SEO content in the early 2000s, and was an easy way to determine how effectively a particular page targeted a search term. SEOs argued about ideal keyword density, with some suggesting 10% was optimal, while others claimed that anything above 2% would be flagged as spam. SEO content writing was just a matter of publishing a page with your desired keyword included enough times, with no real concern for quality of content.
Topic Modeling and Latent Dirichlet Allocation (LDA)
Because keyword density as a ranking factor produced spammy results, this ranking factor was abandoned in favor of other on-page signals. Topic modeling, also known as ‘latent dirichlet allocation’, represents an improved method of determining on-page relevance. Topic modeling / LDA takes the context of keywords into account, and as a result, is better able to understand the relevance of content.
As an example, the keyword “boxer” could have three different meanings: a breed of dog, a kind of undergarment, or a fighter. Google uses topic modeling to refine results so that a search for “boxer briefs” will ignore pages that are about fighters or dog breeds. Only pages related to the theme of ‘apparel’ will be included in the results, increasing overall relevance of searches. Taking this into consideration, it becomes clear that the context of keywords matters – not just the number of times the keyword is used. Because of contextual indicators, the call for ‘good quality content’ is more pertinent than ever before.
Links, Mentions, and Social Signals
Links were originally the part of Google’s algorithm that set it apart from traditional search engines, which acted more like directories. Webmasters would submit their site to the assortment of ‘search engines’ and pray that the meta keywords and meta descriptions would carry them to the top of the results. Google, on the other hand, used inbound links to websites as a ranking factor. The more inbound links, or ‘votes’, a site had, the more authoritative it was viewed by others. This democratic approach to the web produced relevant results and a winning formula that other search engines quickly imitated.
Introduction of rel=canonical
The web’s journalistic integrity was quickly corrupted by spammers who understood the value of links. Spam content marketers could easily manipulate Google’s algorithm by linking to their content on user-generated pages and blog comments. This prompted Google to introduce new ways of combating these spammy links – most notably, the rel=”nofollow” attribute. rel=”nofollow” was meant to discredit spammy links, and helped bring down the amount of comment spam being circulated around the web.
Social Signals: The New Frontier
Matt Cutts, head of the Google Web Spam Team, has hinted at Google’s attempts to understand social media, and how to best incorporate metrics (such as Facebook Likes and Twitter links) in a way that improves the quality of search results. With the new Google +1 feature being rolled out over the coming weeks, Google is looking for ways to introduce social signals and shared links into its algorithm.
Moving Forward with ‘Next Practices’
SEO is not static: It will continue to evolve as Internet user behavior changes and search engines learn to adapt. To maintain and implement good quality SEO, search marketers need to stay current with trends occurring not only within search engine algorithms, but patent registrations, search engine market share, and social media, as well.