Keywords 201: Tails, Association, Authority, and Cannibalization

In my last post, I introduced you to the basics of keyword research. This included the essential pillars of difficulty, volume, and segmentation. You'll need that backstory because today I'm going to build on those concepts to create an intermediate-level treatment of keyword research.

My aim here is for you to come away with a thorough understanding of how keywords relate to one another.

You'll need that knowledge in order to start thinking of your site and your traffic holistically, rather than simply as a mish-mash of posts addressing a mish-mash of keywords (searcher questions).

Keyword Relationships at a Glance

If you're interested in keyword research and haven't yet encountered this head-exploding moment, you will. Let me help you get there.

Here's a quick glance at what ahrefs thinks DaedTech's top performing blog posts are, in terms of search traffic.

Over on the right, you can see the "best" keyword for each of these blog posts, along with the monthly search volume and position in search results. So, for instance:

  • 250 Canadians per month search for "software consultation" and a DaedTech article shows up as the #1 result for that search.
  • And 150 Americans per month search for "expert beginner" and a decade-old rant of mine comes up for that one.
But if I drill in on the top post -- the one about software consulting -- here's what I see in ahrefs.

That blog post ranks on the front page for a whole host of keywords: software consulting, business software consulting, what is software consulting, independent software consultant, etc. And that fact, in turn, will lead anyone in their keyword research journey to ask the question emblematic of the conundrum of keyword relationships.

How do I know which keyword(s) to target with which content?
This core question might manifest in slightly different ways. Sometimes people ask me whether they should "go wide or go deep" with each post, when it comes to keywords. Sometimes people ask whether they should keep creating content about the same topic. But whatever they ask, it always boils down to matching content with keywords.

And I'm going to spend the rest of this post equipping you to do exactly that.

Keyword Length: Long Tails and Short Tails

Before we go any further, we need to understand something about keywords themselves. We need to understand how specificity relates to keywords.

To do that, let's consider a series of increasingly specific keywords, along with their search volumes:

  1. devops (550,000)
  2. devops tools (27,100)
  3. best devops tools (590)
  4. what are the best devops tools (0, reported)
If you were to repeat this exercise with a bunch of single term keywords, adding modifiers and recording their volume, the graph of that data would start to look holistically like this. At least, loosely. That's just me graphing 100,000/x in Google Sheets.

This visual aid gives us the SEO terms "short tail" and "long tail" keywords. So-called "long tail" keywords, reside in the "long tail" of that graph, where more specificity makes the search volume asymptotically approach 0. So I suppose the opposite, "short tail," has come to mean the shorter keywords with higher search volume.

The etymology here isn't super important. We're mainly interested in two important concepts:

  1. Volume is inversely proportional to specificity
  2. Specificity ("long tails") often arises from modifying more basic terms ("short tails")
Not only does this demystify the long/short jargon, but it also informs how keywords relate to one another.

Keyword Relationships

Let's go through those relationships, both defining them and showing real world examples.

1. Parent-Child Keywords

Above, with DevOps, I created a series of parent-child keyword relationships. Each keyword is the parent of the line below it, meaning "devops tools" is the child of "devops." Loosely speaking, child keywords contain their parents.

But it goes beyond that.

Talking about a child keyword would make sense in a post addressing the question implied by the parent keyword. If you were going to write a post that was a definitional guide to DevOps, talking about DevOps tools would make sense. Likewise, in a post about DevOps tools, it would make sense to talk about which tool you thought was "best."

For a real-world example, consider this post from Architect, about staging environments. It addresses the keyword "staging environment" and it contains a section entitled "Best Practices for Staging Environments," which addresses the keyword "staging environment best practices."

This post is well suited to rank for both of these terms. And, just as importantly, it addresses both terms with a single, valuable, coherent piece of content.

2. Keyword Synonyms

If you were watching alertly above, you might have noticed that I abbreviated "best practices for staging environments" to the slightly shorter keyword "staging environment best practices." Does that mean that the latter is the parent of the former?

Nah.

Those two are what I generally think of as synonym keywords. Synonym keywords are slight variations in phrasing around the searcher asking the same basic question. In the case of our two keywords here from the Architect post, the question is the same: what are some staging environment best practices?

So why did I change the keyword at all when explaining it?

It's simple: habit. The more abbreviated version will usually have higher volume, which makes it the alpha-synonym, if you will. More on this later, but that's usually the one to target with content (or a heading).

Lastly, you might wonder if children and parents be synonyms?

Yes, they can. In fact, the Architect post also demonstrates this.

Both the parent keyword "staging environment" and the longer-tail "what is a staging environment" have the same search intent and searcher question, making them synonyms. And the Architect post astutely targets both in one fell swoop.

3. Parameterized Keywords

A proper SEO term for this might exist, but I'm going with a computer-science-y one instead.

I'm going to call this parameterized keywords (in a nod to parameterized unit tests).

Specifically, this is a set of keywords marked by the same phrase, but varied over a single word. For instance, consider this Enov8 post, entitled Data Compliance in Healthcare. One could generalize this to "data compliance in x" and target a group of highly related keywords.

In this case, "data compliance in healthcare" has modest, but existent volume, as do related terms such as "data compliance in nursing" or "data compliance in cybersecurity." So a site could theoretically target these with separate pieces of content or, conceivably, a broad treatment of data compliance across industries, with H2s addressing each industry. Both of these approaches would address the real searcher question, which is essentially "what is the relationship between data compliance and X?"

More on when to do which later.

4. Adjacent Keywords

Finally, let's talk about the easiest conceptual relationship between keywords, which I think of as keyword adjacency. Here, quite simply, we have keyword-questions that might interest the same searcher in the same site visit.

Circling all the way back to the DevOps keywords, think of DevOps tools, with a question of "what are some DevOps tools?" Terms like "devops best practices" or "devops tutorial" might represent adjacent questions. You could reasonably conclude that someone shopping for DevOps tools might also want to know what some DevOps best practices are or how to get going with DevOps, in some fashion.

Topical Authority

I deliberately saved adjacent keywords for my last relationship, for the sake of a nice transition into talking about topical authority. So to make good on that pre-planning, let's talk about topical authority.

The more you address a given topic on your website, the more authoritative the search engine considers you on that topic.

I don't imagine any of you are lighting up CNN's tip line right now with that breaking news. But there is a bit of subtlety here.

Topical authority creates a whole (likelihood of ranking) that is more than the sum of its parts. So targeting 10 adjacent keywords will, all else being equal, yield more traffic than targeting 10 statistically identical, but unrelated, keywords.

This ranking advantage has given rise to a tactic with a pair of relatively tiresome names: topic clusters or hub and spoke content. All this means is that you take a core, shorter tail keyword (like "devops"), target it, and then target child keywords (e.g. "devops tools" and "devops best practices") with other content, preferably inter-linking it all.

But however you assemble your parent-child and adjacent keywords, the core idea remains the same: answer a lot of searcher questions about the thematic topics for which you want to rank.

Cannibalization

So far, I've laid the groundwork for you to absolutely carpet bomb the internet with semi-redundant content. Find parent-child keywords, synonym keywords, parameterized keywords, and adjacent keywords and absolutely firehose at them with content.

Target each individual one with a single piece of content. If that doesn't work, try creating individualized content. Maybe target the same keyword over and over with multiple posts to show you're really excited about that one.

Except... here be dragons. Don't do that.

The search engine's idealized view of your site resembles that of a wiki. You should have a single URL that addresses each individual searcher question. You should then link inter-related pieces of content and define parent-child relationships when elaboration makes sense.

If, instead of structuring things that way, you target the same keyword with multiple pieces of content, you start to cannibalize your content. At the risk of anthropomorphizing the search engine, it looks at your multiple pages that seem to address "What is DevOps" and it says, "I dunno which of these is your final answer, and until you straighten them out, I'm not ranking any of it." Or, worse, it thinks you're a scumbag SEO and are spamming to rank.

So whatever you do, make sure that you have one and only one page on which you plan to primarily address a searcher question.

Heuristics for Planning Content

Now that I've laid out a good bit of keyword and search engine theory, I'll close out by offering some concrete, intermediate-level advice for planning batches of content or entire sites.

Deciding Whether to Target a Parent or a Child Keyword with a URL

Should you write the "(What Is) DevOps" blog post, or should you go straight for "(What Are Some) DevOps Tools" or "(What Are Some) DevOps Best Practices?"

Your first heuristic for evaluating this should be the difficulty and your site's projected rank, given your domain authority. Is the keyword devops anywhere close to attainable? If not, target longer tail, more attainable searches. You can always target the parent keyword later, when you've built some domain and topical authority.

Assuming both are attainable, you should check for cannibalization, which can occur when the parent and child are synonyms. A quick and dirty way to do that is to look at the SERPs for both terms and see if the same articles appear. When that happens, you risk cannibalizaing.

But even absent duplicate SERP entries, you can decide whether you think parent and child keywords both serve the same fully formed question. If they do, then you don't want two separate posts.

Interlinking Parents and Children

Another possible stumbling block is how much to cover in each post when you content separately targeting parent and child. Can you talk about DevOps tools in your "pillar" DevOps post?

If you feel relatively confident in your decision to have separate pieces of content to address the keywords, I'd suggest including a section in the parent post that briefly addresses the child keyword but linking to the fully formed child post for more information. That sends a nice signal to the search engine that, even though the parent post touches on the topic, you consider the link target post to be your canonical take on the matter.

Surround Difficult Terms that You Need to Own

If you're really dying to show up in a SERP but it seems unattainable, target a lot of longer tail keywords related to the SERP you covet.

For instance, take DaedTech, and let's say I wanted to show up for software consulting (and didn't already). Maybe it was important to me because I wanted people executing this commercial search to call me when they wanted software consulting.

I could create a big list of child terms, like this.

I'd then just treat the easy ones as a content creation checklist, building topical authority by "surrounding" the real target. This is a GREAT tactic for service-based business or anyone looking to get in front of searchers at a very specific time.

Target the Alpha Synonym

Earlier, when talking about Architect's staging environment post, I talked about the alpha synonym. At a glance (because I've been doing WAY too much of this the last 5 years), I knew that one keyword would have more volume than the other.

That's the one to target. At least, usually.

When doing your research, capture groups of synonyms, and look for favorable volume and difficulty stats. The one with the most favorable is your alpha, unless it's a very awkward phrasing. Don't Tarzan your readers because the 'best' keyword is awkward.

Pick the best natural sounding keyword to target. The rest will come along for the synonym ride.

De-Index Cannibalization Risks

Let's say that you're cruising along, doing well with your site. But you just can't resist sharing your hot take about what DevOps is going to look like in 5 years. Maybe you title it something like "What is DevOps: Notes from Future Me!"

That's potentially an editorially interesting title. And if your blog has followers, they might click on it to read your take. But to the search engine, it's just a redundant-if-weirder answer to the question "what is devops?"

This doesn't mean you can't publish this content. Just de-index it. Tell the search engine not to bother with it, because you're not intending this content for that purpose.

I would generally suggest de-indexing any content that isn't intended to answer a search question (i.e. rank for a keyword).

Plan Ahead and Track

You'll need some means of mapping out your content and its keywords. And you'll absolutely need to keep track of what you've fulfilled.

This spreadsheet or database should include each URL's primary target keyword, and any children you want to target. It should also have some means of sketching out the interlinking between content and letting you visualize the larger structure.

This serves well for planning and avoiding cannibalization. But it also helps you revisit and refresh content periodically, feeding real "production" data about what you're ranking for back into your planning.

Until Next Time

Hopefully some of that is actionable and helpful for you. In the next installment, I'm to build on this and offer some advice on how to construct wholesale campaigns and content roadmaps. And I'll try to do it sooner than 6 months from now.