Last year I audited the blog of a regional plumbing company. They had 68 posts published over four years. Topics ranged from water heaters to bathroom remodel trends to a recipe post someone thought was a good idea in 2022. Total organic traffic: about 400 visits a month. We deleted or merged 30 posts, reorganized the rest into four tight topic clusters, and interlinked everything on purpose instead of by accident. Nine months later the site sat at 3,100 organic visits a month, and two of the four pillars held top three positions for their head terms.
Nothing about the writing got better. The structure did. That is topical authority in one story: Google trusts sites that cover a subject in full and prove the coverage with their internal links. Small sites can win this game because you do not need a big budget. You need focus.
What Topical Authority Actually Means
Topical authority is not a metric in any tool; it is an outcome. When your site answers most of the reasonable questions searchers have about a subject, Google starts treating your domain as a reliable source for that subject. Rankings get easier across the whole topic, not just on individual posts. New posts in the same topic index faster and rank higher out of the gate.
Two things create that outcome. First, coverage depth. You have addressed the head term, the subtopics, and the long tail questions around them. Second, internal linking structure. Your links show Google how the pieces relate: which page is the pillar, which pages support it, and which siblings answer adjacent questions. Coverage without linking looks like a pile of posts. Linking without coverage looks like a doorway page. You need both.
Google has been explicit for years that it rewards helpful, people-first content and demotes thin content made for search engines. The creating helpful content documentation says it plainly: demonstrate real experience and satisfy the searcher’s need. A cluster does exactly that at the topic level.
| Approach | What you ship | What Google can infer |
|---|---|---|
| Scattered blogging | Random topics, weak or accidental links | A pile of unrelated pages |
| Doorway linking | A hub that points at thin spokes | A directory with no real answers |
| Topic cluster | One pillar, 11 to 15 intent-specific spokes, descriptive anchors both ways | A subject covered in full |
Step One: Build a Topical Map Before You Write Anything
A topical map is the full list of pages you would need to cover a subject end to end. Building one takes a day and saves you months of wasted writing.
Start with one seed keyword tied to revenue. For the plumbing company it was “water heater repair.” Not “plumbing,” which is too broad, and not “wedding venues,” which is what their recipe post was accidentally competing for. One seed per cluster.
Expand the seed three ways. First, Google itself: autocomplete and the People Also Ask boxes give you the real questions searchers type. Second, competitor gaps: take two or three competitors that outrank you, list the subtopics their top pages cover, and note the questions none answer well. Third, your own inbox. Customer emails and sales calls are long tail keywords with the wording already done.
Then group everything by intent. “Water heater repair cost” and “how much is a new water heater” belong on the same page because they share intent. “Water heater repair cost” and “signs your water heater is failing” do not, even though both mention repair. Skip the intent grouping step and you build cannibalization into the cluster on day one. Our search intent analysis framework walks through that grouping process in detail.
You should end with 12 to 20 page ideas for a first cluster. If you end with 80, your seed was too broad. Split it.
Step Two: Ship the Pillar and Clusters in a Tight Sequence
Here is the actual 12-post cluster we shipped for the plumbing company, because abstract advice is useless without a shape to copy.
pillar: water-heater-repair-guide/ (links down to all 11 cluster posts)
cluster posts (11): tankless-vs-tank, repair-cost, signs-of-failure,
maintenance-checklist, permit-rules, gas-vs-electric, lifespan,
emergency-leak, brand-comparison, installation, troubleshooting
each cluster post: links up to the pillar in the first two paragraphs,
links sideways to 2-3 siblings where genuinely relevantThe pillar is a 2,500 word page that touches every subtopic in a paragraph or two, then hands off to the cluster post for the full answer. It targets the head term; every cluster post targets one specific intent at 900 to 1,500 words.
Sequence matters. Publish the pillar first, even though it links to pages that do not exist yet, and add the links as each cluster post goes live. Then ship two cluster posts a week for about six weeks. Google sees the topic fill in as one deliberate effort, not random monthly blogging.
Get each new URL indexed the day it ships. Request indexing in Search Console, and if your stack supports it, ping IndexNow so Bing and the other participating engines pick it up immediately; the IndexNow documentation shows the single-URL ping format. Fast indexing is not a ranking factor, but a cluster only works once the pages exist in the index and the links between them count.
Step Three: Interlink Deliberately, with Descriptive Anchors
Internal links in a cluster follow rules, not vibes.
The pillar links to every cluster post. Every cluster post links back to the pillar, high on the page, so the hierarchy is unmistakable. Cluster posts link sideways only when a reader would genuinely want the sibling next: the repair cost post links to the brand comparison post, not to the permit rules post just because it exists.
Anchor text does real work here. “Click here” and “read more” waste the signal. The anchor should describe the destination page the way a searcher would: “water heater repair costs” or “tankless versus tank water heaters.” Descriptive anchors are one of the cheapest relevance signals you control entirely, and Google’s crawlable-links guidance treats anchor text as a signal about the destination. Nobody can take them from you, and no algorithm update has ever penalized a site for clear internal anchors.
Two habits keep the structure honest. Every new post ships with three to five internal links out. And when a new post goes live, you spend fifteen minutes adding two or three links into it from older posts. Links only into new content and never back out of it is the single most common cluster failure I see.
Step Four: Prune and Merge the Overlap
Clusters decay. Six months in, two posts start competing for the same query, or a post targets something nobody searches, or a product change makes a post wrong. Left alone, overlap drags the whole cluster down because Google splits signals between the competing pages and ranks neither.
Every quarter, run a pruning pass. Pull up Search Console and look for queries where two of your own URLs swap positions week to week. That is cannibalization, and the fix is a merge: fold the weaker post into the stronger one, 301 redirect the old URL, and update the internal links to point at the survivor. Posts with zero impressions after twelve months get the same treatment: fold them into a sibling or delete them outright. Fewer, stronger pages beat more, weaker ones almost every time.
The plumbing company’s 68 posts became 38, and traffic went up eight times. Deleting content feels wrong until the first time it works. A full pruning pass pairs well with the technical checks in our technical SEO audit checklist, since redirect chains and orphaned pages tend to hide in the same corners of a site.
Step Five: Measure Query Coverage, Not Just Traffic
Traffic is a lagging indicator for clusters. The leading indicator is query coverage: how many distinct queries your cluster ranks for, and how that set grows as you ship.
Open the Performance report in Search Console and filter queries containing your seed term and its close variants. Export the list. Count the distinct queries where any page in the cluster has impressions. That number is your coverage. When we started, the water heater cluster showed impressions for 19 queries. Six months later it was 340. Traffic followed coverage with about a two month delay, which is the normal pattern.
Check coverage monthly, not daily. Daily checking will make you crazy, because a new post often sits at position 40 for weeks before it jumps. What you want to see is the query count climbing steadily and average position for the cluster’s queries drifting from the 30s into the teens. If coverage is flat after 90 days and the cluster is fully shipped, the problem is usually search demand or competition, not your structure. The Search Console performance report documentation covers the filtering and export options if you have not used them beyond the default view.
One more reason to care about coverage: AI search engines cite sources that demonstrate depth on a topic. The same cluster structure that builds classic topical authority also makes your site a better citation candidate, which we cover in answer engine optimization. You get both payoffs from one structure.
When Topic Clusters Are the Wrong Move
I recommend clusters often, but not always.
If your business is one service in one town, a cluster may be overkill. A plumber who only does drain cleaning in one city needs five strong service pages and a Google Business Profile, not a 15-post content operation. Clusters pay off when the topic has real query depth and your buyers actually research before they call.
If you cannot commit to the full sequence, do not start. A pillar with two cluster posts is not a cluster. It is three orphaned pages with extra steps. Half-built clusters underperform no cluster at all because you spent the budget and got none of the compounding. Finish one cluster before starting a second.
And skip topics with no line to revenue. The recipe post on the plumbing blog ranked fine for a while. It generated zero calls. Every cluster should answer the question: if this ranks, does someone become a customer? If the honest answer is no, pick a different seed.
What Most Guides Get Wrong
Most guides on this topic start with keyword research tools and end with a diagram of a hub and spoke. That is backwards, and it is why so many cluster attempts die at three posts. The tool-first approach treats topic selection as a volume problem when it is really a depth problem. A seed with 40,000 monthly searches and thin subtopic coverage loses to a seed with 2,000 searches and twelve well-answered questions every time. Guides also treat the pillar page as a long blog post. It is not. It is a routing page that earns its word count by summarizing every subtopic well enough that readers and crawlers know exactly which spoke answers which question. Another common miss: publishing order is presented as flexible. It is not, because a pillar published after its spokes has no historical internal links pointing at it, and you end up rebuilding anchors across a dozen live pages. Guides also skip the pruning half of the job entirely, as if clusters only grow. Every cluster I have maintained for more than a year needed at least one merge, and pretending otherwise leaves Google splitting signals between two competing URLs. Finally, almost nobody says this plainly: a cluster underperforms a single great page when the topic has shallow demand. The checklist mentality says build it anyway. The honest answer is to check query coverage data first and pick a different seed if the depth is not there.
Conclusion
Topical authority is coverage plus structure. Build a topical map from one revenue-tied seed keyword, ship the pillar first, then two cluster posts a week until the map is done. Interlink with descriptive anchors in both directions, prune overlap every quarter, and measure query coverage in Search Console instead of watching daily traffic. Give the cluster 90 days before you judge it. One cluster of 12 to 15 posts, fully shipped and fully linked, will outrank a hundred scattered posts on the same subject. Pick your seed, map the topic, and start shipping.
