When SEO was first emerging as a strategy, there was one consideration more important than any other: keywords. First, you’d do exhaustive keyword research to see where you ranked versus a leading competitor, and evaluate how much traffic those keywords received. Eventually, you’d identify a handful of “target” keywords to use in your campaign (picking ones with the highest search volume and lowest competition rating), and start injecting those keywords into your meta data, body content, and external links.
Now, there’s a new school of thought with regard to keywords, thanks to Google’s semantic search capabilities; it’s better to select key topics, rather than keywords specifically, to avoid the temptation of over-optimizing or “keyword stuffing” and go after niche opportunities that address consumer inquiries.
But what exactly is the difference between keywords and topics, and which one is more important if your end goal is search engine visibility?
Welcome to Hummingbird.
Have you ever typed a long question into Google, expecting to wade through dozens of results but instead finding the exact resource you needed? You have this guy to thank. Google Hummingbird is a semantic search algorithm, which evaluates the intention behind a user query rather than trying to match keywords to pockets of similar keywords on the web. It may be an oversimplification to say it “understands” what you’re searching for, but that’s pretty close to the truth.
Allow me to demonstrate this with the search query “how to find a good suit”:
Notice there aren’t any top results with the exact phrase “good suit” or “find a good suit,” which we could take to be the core keywords in this query. Instead, Google has evaluated my intention: figuring out how to buy a good, well-fitting, general purpose suit, and has come up with options like “How to buy a suit,” and “how should a suit fit? Your easy-to-follow visual guide.” These topics don’t match my query keyword-for-keyword, but they serve this function and therefore rank.
This is only one example, but it illustrates my point that good topics have a stronger tendency to rank than topics with specifically selected keywords.
It’s important to consider the fact that while not always the case, striving to include more keywords often leads to inferior content from a user experience perspective. Stuffing keywords into your pages, blogs, and body content will come across as unnatural, and probably won’t serve your users’ needs with any degree of utility. Choosing topics, then, is a way of writing for users more than search engines, which does come with advantages and disadvantages. As a general rule, I side with user experience; I wouldn’t stand much to gain by ranking if my user experience left a bad impression.
Again, this is a general rule, so bear that in mind. When choosing topics, you usually opt for highly specific, long-tail phrases over shorter, more common keywords. That means you’ll likely have a higher relevance and a faster path to the top with a topics, but much lower search volume.
Note the difference between “avocado recipes” and “how to prepare an avocado”:
The difference is dramatic. (Also note how popular avocados have become since 2009).
Search volume isn’t everything, but it is an important consideration.
Just because Hummingbird is designed to understand search queries doesn’t mean it’s always good at it. For example, let’s say I’ve searched for “avocado recipes” as in the above illustration:
The results, as you might imagine, are avocado recipes—and only recipes. There are no topics or articles that explain why avocados are a good ingredient, or what types of dishes they’re used in, or how to prepare them once you’ve got a recipe in mind. In this way, you can’t always be certain that your choice of topics will appear for your intended range of queries.
Going along with this, keyword-based SEO strategies have always been rooted in data; you evaluate the search volume, monitor the competition, and calculate a kind of cost-to-benefit ratio before deciding which ones to move on. Topic selection is a much more qualitative approach; you might conduct surveys and evaluate the past performance of your content, but there isn’t much hard data to back up your assumptions. In marketing, assumptions are almost universally bad, so unless you have hard data to guide you in your choice of topics, you stand to lose significant potential.
So far, I’ve only discussed keywords as they relate to ongoing content. When it comes to site pages, title tags, meta descriptions, and even anchor text for your links (though you must be careful not to over-optimize your anchor text), keywords play an important role. Additionally, these applications carry less of a user experience detraction than their application in content alone. Thus, keywords are still better than general topics in at least some applications.
It’s hard to come to a concise and satisfying conclusion on this matter, because it really is a complicated subject. Still, we’ve seen the evidence and can compile it into a handful of key takeaways:
Accordingly, I think keywords are still important—but they should only serve as guides to help you come up with great topics, and as anchor points to help guide search engines in the right direction. In short, do your keyword research before selecting your topics, but don’t sacrifice your direction or content for a handful of keyword phrases.