The Search Engine Optimization industry is abuzz with all sorts of metrics — Vol, CPC, CI, KEI, NPR (ok, ok, the last one wasn’t a metric). Technical marketers love to slice and dice numbers to find those diamond-in-the-rough keywords for their SEO campaigns. And rightly so! One of the best aspects of performance marketing is that it’s quantitative and measurable, great news for analysis-oriented left-brain types.
But there’s one metric that’s always been missing from every SEO’s formidable arsenal. It’s the metric that everyone talks about, marketers aspire to achieve, but (so far) has been notoriously difficult to quantify. And that metric is: Relevance.
Industry experts, and Google itself, urges marketers to write high-quality content and to focus on your target audience — but identifying which topics are relevant to your target market is left up to you. Google is certainly calculating relevance and scoring it in a quantitative manner; these relevance calculations are core to determining Topical Authority, PageRank, which synonyms are matched by their internal engine and more. But they are not providing the relevance data back to marketers, so as not to tip their hand. Without that relevance data, however, marketers simply lack the necessary information to make data-driven decisions on the relevance of their content.
Without Google giving us clear directions on how to model relevance, it’s been up to us marketers to find workarounds. As pointed out by Moz, one good method of determining semantic relevance is to build a topical hub, using some relevance-based data source such as Wikipedia, and to then look at entities to determine which is the most relevant. This is certainly a valid way to do it, since Wikipedia is bound to have high-quality, relevant content for basically anything you’d want to market. As you’d expect, however, it’s difficult to execute this at scale — marketers have to manually build topical hubs for each topic that they’re marketing, and for each client, which can quickly get out of hand.
SEO tools today also fail to provide relevance-based metrics. Great keyword research tools such as SEMRush and SpyFu return keywords that have some degree of relevance but they can’t quantify the degree of relevance, so their data can only be sorted on Volume. For example, at the time of this writing, SEMRush will generate 10,000 keywords for each query. But once you have the list, it’s up you to manually determine relevance for each topic — and combing through 10,000 keywords for each topic is no small task.
Another issue has to do with the way that the ‘relevant’ keywords are generated. The easiest way for a keyword tool to generate suggestions is to look at the seed term you’ve entered (e.g. “canned dog food”) and expand it to a list of niche keywords that contain the seed (e.g. “best canned dog food for senior dogs”). Long-tail keyword campaigns have always relied on this type of expansion. However, as Google moves toward semantic search, it’s begun making adjustments to the way it lists Search Engine Result Pages (SERPs): Google is now lumping results, meaning that it shows the same SERP for many long-term keywords. As a result, there is less of a discernible difference between ranking for “canned dog food” and “best canned dog food for senior dogs”.
We started MarketMuse in 2013 to bridge this gap and to provide marketers the relevance data that they need to be effective. Working with a team of PhD statisticians, we’ve built the MarketMuse Keyword Relevance Engine, the core technology that powers our set of content analysis tools. MarketMuse differs from any other keyword research / content analysis tool in two important ways:
1. MarketMuse generates topically-related keywords which don’t contain the seed term. As an example, our technology draws the connection between “dog food” and “pet food”. We’ll serve both search synonyms (‘proof terms’) and related keywords (‘related terms’), depending on which terms will rank you highest in organic search. (For those who have read Cyrus Shepard’s great blog post on the Moz blog, think along the lines of TF-IDF and co-occurrence measures).
2. For each related keyword, we calculate a Relevance score, which measures the degree of topical relevance. This relevance score is critical to prioritizing how you spend your time. Instead of having to comb through thousands of keywords, MarketMuse just tells you the most important keywords you should focus on right now.
When you create content that draws a relevant audience, that content is much more likely to drive real results to your bottom-line. A central question in organic search marketing is conversion: you’re drawing pageviews on your website, but how many visitors will convert to customers? By prioritizing your content creation on Relevance, you’re more likely to draw the right audience that will actually buy your products and services.
NOTE: The MarketMuse Blog Analyzer was used to optimize this post for “seo keyword relevance” and “keyword research”.