SEO company SearchMetrics has just released it’s 2014 Ranking Factors Study, an analysis of what factors really matter in determining where your webpage ranks in search engines. And the results are in: Relevant Terms are the #2 ranking factor for determining where your webpage ranks on Google (second only to historical Click-Through Rate). This should give pause to both dyed-in-the-wool SEO experts and marketers completely new to web content, as it heralds a significant shift in your SEO strategy.
Since Google launched its Hummingbird search engine update in 2013, the landscape of what matters in search engine optimization and content marketing has shifted. In the past, the main factors for determining Google Page Rank were largely a combination of backlinks (i.e. which inbound links are pointing to your page), technical SEO factors (i.e. factors such as whether your title and header tags correctly reflect the keywords you want to rank for) and, more recently, social signals (i.e. whether people in your social network have endorsed this content). But the topical nature of Hummingbird — coinciding with the steady rise of content marketing — changes everything.
Instead of looking at the SEO in your META tags or other formatting elements within your webpage, Hummingbird leverages a new set of patents and the Google Knowledge Graph to analyze the topics covered in the body of your content. Effectively, this is a shift from traditional format-based SEO to more organic SEO topical authority. But how does it do this?
First, the low-hanging fruit with regards to relevant terms is synonym matching. As SEO by the Sea points out, Hummingbird utilizes a synonym matching engine that identifies synonym terms, also known as substitute terms (e.g. “car” and “auto”), that searchers may use interchangeably in a search query. This is similar to how Google’s broad match and phrase match already works: by identifying slightly different search phrases that have a common search intent, Google can drive traffic to the best sites for that intent, not just for the wording of the search phrase.
Secondly, there are topically-related terms. Related terms are topics that aren’t synonyms but are still conceptually-related. For example, “dog food”, “pet food” and “doggy treats” are topically related keywords, but they’re not exactly synonyms of each other. By analyzing very large quantities of web content, and looking for the co-occurrence of topics as well as other factors, Google (and MarketMuse) are able to determine the relevant terms (what we call “related keywords”) for a given keyword. Although Google doesn’t expose this data, MarketMuse tools do let you check your content to identify topical gaps that are preventing you from ranking higher in organic search.
As a third category, there are semantically-related terms. These are topics that have a connection within language based on their meaning. For example, a “dog” is an “animal”, so you would expect a topical connection between these two keywords reflected in content as well. Often, semantically-related terms can be subtopics within a topic or a characteristic of a topic. The promise of a truly Semantic Web has been a long time coming, but due to the difficulty of accurately modeling semantic relationships, this category has seen less momentum in practice. That said, repositories such as WordNet at Princeton and ConceptNet at MIT do exist and have been used heavily in Natural Language Processing research to further advance this area.
What we do know is that the push for “high-quality content” will continue. Content marketing ups the ante for successfully writing quality content, both in terms of novelty, readability and pure SEO value. Marketers that get this right will find that their search engine rankings have improved by leaps and bounds, while marketers that fail to write authoritative content will see their content left by the wayside.
NOTE: the MarketMuse Blog Analyzer was used to optimize this post for “seo related keywords” and “topical authority”.
Featured image vector designed by Freepik