2 kinds of relevance
My colleagues Frank and Cynthia are at SES in New York while I sit jealously connecting with them from my desk, and reviewing and approving the typeset pages for our book. Expect a repcap of the conference at the end of the week in this space. In the meantime, I’ll share with you some fresh perspectives on the book’s content gleaned during the review process.
We finished the writing process last September. But this is the fourth opportunity we’ve had to review and edit drafts. Each time I see the book fresh, I get more excited about it. I’m not just saying this out of self interest: But I am really proud of this book. I think it will make a big difference for writers, editors, and content strategists.
Like much of the book, the fresh perspectives I want to share in this post are about content relevance. Ours is the first book-length effort I know of to treat the subject in a rigorous way as it applies to Web content. Most advice I have read about Web content relevance either assumes everyone knows what relevance is or is inaccessible to the people who create and manage content for the Web. Our goal was to build a model of relevance that is both rigorous and accessible. After reviewing the typeset pages, I think we’ve accomplished that goal.
Central to our model is a distinction between two kinds of relevance: semantic and contextual relevance. Semantic relevance is all about how well the meaning of the words within pieces of content are related. Contextual relevance is about how well a piece of content is related to the salient circumstances in which it is communicated. Most people understand semantic relevance because they do keyword research. From a search perspective, that’s all about matching your page code to Google’s algorithm. Contextual relevance is less well understood. From a search perspective, that’s all about getting your site interwoven into the fabric of credible sites related to your topics.
If you’re interested in this distinction, and how it affects search and social media, read on.
Before I talk about the distinction, I want to talk about how we define relevance. I’ve already tipped my hand in this blog about how we define relevance, but it bears repeating. We don’t define relevance in terms of the words and phrases people use but in terms of how those words and phrases affect people. For Web users, relevance is a function of two things: How much change a piece of content creates in the mind of a visitor and how much effort it takes the visitor to process that change. A piece of content is relevant to a Web site visitor to the extent that it creates a relatively large change in the mind of the visitor and it requires a relatively small amount of effort to produce that change.
When we think of semantic relevance, a paradigm case is the relationship of keywords to the search engine results pages (SERPs) we see in Google. Google constantly runs A/B tests to see how their test users click search results. In general, content that users click more often climbs the ranking. This is balanced against bounce rates. Content that users bounce from drop in the rankings. Google assumes that if users click links in the SERP and do not bounce, those users find the content semantically relevant to them. That is, the meaning of the keywords is sufficiently related to the meaning of the content to make some impact on the minds of users with little enough effort.
When most people speak of relevance, they mean semantic relevance. That’s its common or conventional use. Contextual relevance goes beyond that. In day-to-day conversation, contextual relevance is all about how two utterances are related. If I say something and you respond, the relevance of your response is related to what I said, not just to the meaning of the words. In face-to-face conversations, we can skip all kinds of words and otherwise abbreviate our utterances because of all the contextual cues–gestures, pre-existing relationships, the history of the conversation, etc. Context greatly enhances relevance.
If we extend the conversational metaphor to the Web, a piece of content is contextually relevant to the extent that it is related to other pieces of content in the greater conversation. Of course, these relationships are just links on the Web. You might write something profound, but if it does not relate well to the greater conversation, it’s contextual relevance is limited. It might still be relevant to a select few. But it misses the opportunity to enhance it’s relevance by connecting with the most credible sites on the same topic.
Contextual relevance is measured by Google in terms of PageRank. The more links between your content and other credible sources on the Web (especially from credible sources to your pages), the higher the PageRank. PageRank is more than half of the Google algorithm for a reason. PageRank gives contextual relevance its full force in search results.
In most cases, semantic relevance is not enough for Google to recognize your pages by giving them prominence in SERPs. You need contextual relevance to do that. There are no short cuts to doing this. You have to create and promote gifts of knowledge–pieces of content that credible sources want to link to. You have to promote these gifts of knowledge on Twitter, Facebook and LinkedIn. You have to seek out influencers in your topic areas and connect with them on and off your site. And you have to hope your efforts will result in the credibility you need. When you get there, PageRank will follow.
You can and you must try to make your content more contextually relevant. But you can’t try to get PageRank. Google will only reward you with PageRank after your content is deemed contextually relevant.
James Mathewson is the EIC of ibm.com and the Search and Social Media Strategy Lead for IBM Digital Marketing.