You have 37 blog posts. You published them over the last year and a half. Each one targets a different keyword. Some of them are pretty good. A few even rank on page two for something.
But your traffic graph looks like a flat line with occasional hiccups.
This is one of the most common patterns in content marketing, and it was captured perfectly in a recent Reddit thread where an SEO practitioner laid it out: "If your website talks about everything, Google assumes you're an expert in… nothing."
The thread blew up — not because the insight was revolutionary, but because nearly every commenter had lived it. One agency owner described clients who wanted 50 articles across random topics and ended up with a blog that ranked for nothing. Another commenter explained how they stopped chasing volume and started thinking in clusters, and "Google noticed, but AI search noticed even more."
The consensus was clear: connected content beats scattered content. But knowing that and actually doing it are different problems. Most people get stuck in the gap between "I have keyword ideas" and "I have a coherent plan." That gap is where topic clusters live — and where most small sites never build the structure that makes compounding possible.
What a Topic Cluster Actually Is (and Isn't)
A topic cluster isn't a fancy content calendar. It's a structural decision about how your site's content relates to itself.
At its simplest: you pick one core topic your audience cares about, then write 8-15 posts that explore different angles, questions, and subtopics within that space. Every post links to the others where it makes sense. Together, they tell Google — and your readers — that you actually know this subject deeply.
The opposite of this is what most blogs look like: a post about email marketing, then one about pricing strategy, then one about Instagram Reels, then one about client onboarding. Each post exists alone. Nothing connects to anything else. A visitor who reads one post has no obvious next thing to read.
One commenter in the thread nailed the practical filter: "If someone reads this, is there a clear next question we answer? If not, probably not worth writing."
That's the whole idea. A cluster isn't built around keywords that happen to have high search volume. It's built around a reader who has a problem and needs to understand multiple dimensions of that problem before they can solve it.
If you're a coach who helps people with productivity, one cluster might cover everything around building a weekly planning system — why people fail at time management, how to audit where your time goes, tools for weekly planning, how to handle interruptions, what to do when the system breaks down. Each post answers a different question, but they all serve the same person on the same journey.
How to Pick Your First Cluster
This is where most people stall. They know clustering matters, but they can't decide where to start.
Here's a straightforward approach: look at what you've already written.
If you have 20+ published posts, you probably already have the seeds of two or three clusters buried in your archive. Pull up your full list of posts and sort them — not by date, but by topic. Group anything that could plausibly be read by the same person trying to solve the same type of problem.
You'll likely find one group that's bigger than the others. That's your first cluster candidate.
If you're starting closer to scratch, pick the topic where these three things overlap:
- You have genuine depth. You could talk about this for an hour without checking notes. This is where your actual expertise lives, and it's what separates your content from the AI-generated summaries flooding every niche.
- Your audience asks about it. Not "a keyword tool says it has volume," but real people in your DMs, on your intake forms, or in your comments are asking these questions.
- It connects to what you sell. The cluster should naturally lead someone closer to understanding why they'd want your help.
That Reddit thread surfaced an important tension: one commenter asked, "How do you balance going deep on one topic vs. staying broad enough that you don't pigeonhole your site?" The answer is that you don't have to pick one topic forever. You pick one topic first. Build real depth there. Then expand to the next cluster once the first one is solid. Pigeonholing isn't really a risk — you're not committing to only writing about email marketing for eternity. You're committing to finishing one coherent body of work before scattering your effort across five.
If you've already done keyword research and have a list of promising terms, the clustering step is where you decide which of those keywords belong together — which ones serve the same reader and the same problem space. That's the strategic grouping layer that turns a keyword list into an actual plan.
Anatomy of a Cluster That Works
A functional cluster has three layers:
The pillar. One comprehensive post that covers the core topic broadly. This is the page you most want to rank. It addresses the main question and points readers toward more specific posts for deeper exploration. Think of it as the table of contents for your expertise on this subject.
The supporting posts. These are the 8-12 posts that go deep on specific angles. Each one targets a more specific keyword or question within the topic. They're the posts that answer the "next question" your reader would naturally ask.
The internal links. This is the connective tissue, and it's where most people drop the ball. Every supporting post should link to the pillar and to at least two or three other supporting posts where the connection is natural. The pillar should link out to every supporting post. These aren't decorative — they're how Google understands that these pages are related, and they're how readers actually move through your content instead of bouncing after one page.
Here's what this looks like in practice. Say you're building a cluster around "creating content for your business website." Your pillar might be a broad guide to what your business website should write about. Supporting posts could cover how to write your first blog post when you're not a writer, how to plan content without a marketing team, how to use AI assistance without losing your voice, and how to outline posts quickly. Each of those posts stands on its own, but together they build a body of knowledge that no single post could achieve.
The linking between them matters more than most people realize. When someone lands on your post about outlining blog posts, they should see a natural path to the post about writing faster or the one about finding the right topics. Not because you shoved links everywhere, but because the reader's journey logically goes there.
As one commenter put it: "I'd stop publishing the moment you can't explain how the new post fits with the last three." That's a high bar, but it's the right filter.
How to Know When a Cluster Is "Done Enough"
This question haunts people, and it shouldn't. A cluster doesn't need to be perfect or exhaustive before it starts working. It needs to be coherent.
Here's a practical threshold: your cluster is ready to start compounding when you can answer yes to these questions:
- Does the pillar post exist and cover the core topic without major gaps?
- Do you have at least 5-6 supporting posts published and interlinked?
- Can a reader land on any post in the cluster and find at least two natural links to other posts in the same cluster?
- Does the cluster cover the main questions someone would ask while trying to understand or solve this problem?
You don't need 15 posts before the cluster "works." You need enough posts that the structure is visible — to Google and to your readers. Five to six well-linked posts around a clear theme will outperform twenty disconnected posts on random topics almost every time.
Once your first cluster reaches that threshold, you have a choice: keep adding supporting posts to deepen it, or start building your second cluster. Both are valid. The deciding factor is usually whether your audience still has obvious unanswered questions within that topic. If you keep hearing the same question and you haven't addressed it, write that post. If the cluster covers the journey well, move on.
The key insight from the Reddit thread — echoed by almost every commenter — is that this feels slower than just publishing whatever keyword comes up next. And it is slower, at first. But scattered content doesn't compound. Connected content does. One commenter described it as "content ecosystems" where "one piece feeds the next, each one gives the reader a reason to keep going deeper." That's the compounding effect. Each new post in a cluster makes every other post in that cluster slightly more valuable.
Reorganize Before You Write More
If you already have a backlog of published posts, the highest-value move isn't writing something new. It's reorganizing what you have.
Go through your existing content. Group posts into potential clusters. Identify gaps — where a cluster has three posts but obviously needs a fourth to address a common question you've skipped. Add internal links between posts that belong together but currently don't reference each other.
This is the kind of work that a content pipeline is built for — not just scheduling new posts, but seeing how everything fits together and where the structural gaps live.
You might find that you only need to write two or three new posts to turn a scattered collection into a coherent cluster. That's dramatically more efficient than starting from scratch, and it gives your existing content a better chance to actually perform.
The thread started with someone describing the common trajectory: find keyword, write article, wait for traffic, repeat 40 times, wonder why nothing's working. The fix isn't to write 40 more. It's to make the 40 you have work together — then add deliberately from there.
Pick one topic. Build the cluster. Link it together. Watch what happens before you move to the next one. That's how content compounds instead of just accumulating.