A content calendar lists what you will publish and when. A content strategy decides which topics you will own, how the pieces relate to each other, and what you will deliberately not cover.
The second is rarer, and the absence shows up predictably: a blog with four hundred posts, no clear topic ownership, several pages competing for the same query, and traffic concentrated in a handful of pieces nobody planned.
Start From What You Can Credibly Own
The first decision is scope, and most sites get it wrong by going too broad.
A site covering thirty loosely related topics has shallow coverage of each. A site covering four topics thoroughly builds the interconnected depth that both search engines and readers recognise as expertise, which is the mechanism covered in the topical authority guide.
The test for scope: would a knowledgeable person in your field consider you a credible source on this topic. If the honest answer is no, publishing on it produces content that competes with genuine experts and loses.
Narrowing is uncomfortable because it feels like leaving traffic on the table. The traffic you leave is traffic you were not going to capture, and the concentration produces better results in the topics you keep.
Clusters, Not Individual Posts
The structural unit is a topic cluster: a comprehensive pillar page covering a subject broadly, supported by focused pieces covering its components, all interlinked.
This works for two reasons. It matches how people actually research, moving from broad understanding to specific questions. And it creates the internal link structure that distributes authority across related pages rather than leaving each post isolated, per the internal linking guide.
Plan clusters before writing anything. Identify the pillar topic, list the component questions, and check which of those questions have enough distinct search demand to warrant their own page rather than a section.
The check for whether two questions need separate pages is SERP overlap. Search both and compare page one. Substantial overlap means one page covering both. No overlap means two.
Match Content Type to Intent
Every planned piece should have a stated intent and a format that serves it. Mismatches are the most common reason well written content fails to rank.
Informational intent wants explanation, and a product page will not rank there regardless of quality. Commercial investigation intent wants comparison, and a general guide will lose to a direct comparison. Transactional intent wants a page that lets someone act.
Determine intent by searching the term rather than reasoning about the phrasing. Whatever format dominates page one is the format Google has decided that query wants, and competing with a different format is a fight against a decision already made.
Decide the Ratio Between Traffic and Revenue Content
Informational content earns traffic. Commercial content earns revenue. Most content programmes drift heavily toward the first because it is easier to write and produces better looking numbers.
The drift is worth catching deliberately. A programme where nine in ten pieces are top of funnel explainers will report growing sessions and contribute little to pipeline, which is the measurement problem covered in the pipeline measurement guide.
My position: for most B2B businesses the ratio should be closer to even than it usually is, and comparison content specifically is underweighted relative to how close it sits to a decision. The comparison pages guide covers why.
Publishing Cadence Matters Less Than Consistency
Frequency is not a ranking factor and it is treated as one constantly.
What matters is whether each piece is good enough to compete for its target query. Four thorough pieces a month beats sixteen thin ones, and the thin ones create the dilution problem covered in the pruning guide.
The cadence worth setting is one your team can sustain while maintaining quality, then holding it. Erratic publishing followed by long gaps produces worse results than steady modest output, mostly because momentum in a topic cluster depends on completing it.
Plan Maintenance as Part of the Strategy, Not After It
Every piece published becomes something that needs maintaining. A strategy producing forty pieces a year is also producing forty future maintenance obligations, and most content plans account for the first and not the second.
Budget for it explicitly. A reasonable split allocates meaningful capacity to updating existing content rather than treating refresh as whatever fits after new production, which in practice means never. The refresh workflow guide covers the operational process.
This constraint should influence how much you plan to publish. A team that can produce eight pieces a month but only maintain twenty total pages should not publish ninety six pieces a year.
Differentiation Is the Part That Cannot Be Systematised
A well structured cluster covering the right topics at the right intent still fails if the content itself is interchangeable with what already ranks.
What makes a piece worth ranking is usually one of a small number of things: original data nobody else has, first hand experience nobody else can claim, a genuine position on a contested question, or specificity where everyone else generalised. The non-commodity content guide covers this in depth.
Plan for it explicitly at the brief stage. A brief specifying a target keyword, a word count, and an outline will produce competent commodity content. A brief specifying what this piece knows that competitors do not will produce something worth linking to.
The Brief Determines the Outcome
Most content quality problems are brief problems. A writer given a keyword and a word count produces content matching that specification, which was the wrong specification.
A brief that works states the specific question the piece answers, who is asking it and what they already know, the position the piece takes, what evidence supports it, which existing pages it links to and which will link back, and what it deliberately does not cover.
That last item prevents scope creep into topics owned by other pages in the cluster, which is where cannibalisation originates.
Measuring the Strategy Rather Than Individual Posts
Per post traffic is noisy and encourages judging pieces individually when the cluster is the unit that matters.
Better measures: whether the cluster as a whole is gaining ranking coverage across its component queries, whether the pillar page is accumulating links, and whether the commercial pages in the cluster are receiving internal link equity from the informational ones.
Search Console filtered to a URL path gives you cluster level performance directly, which is more useful than any individual page's numbers.
What This Also Builds
Cluster structure, clear intent matching, and genuinely differentiated content are the same foundations that determine whether AI systems can find and cite your content. The conversational search guide covers where the two diverge, which is mainly at the research layer rather than the structural one.
Practically, that means a content strategy built properly for search does not require a parallel strategy for AI visibility. It requires the same strategy executed with the technical foundations in place.
Checking Coverage Gaps
The NotionCue AI Topical Cluster Map shows where your coverage is genuinely deep and where it is thin, which is the input a cluster plan needs and the thing most content audits approximate badly.
The NotionCue AI Answer Gap Finder surfaces which sources currently answer questions in your topic area, which tells you what a competitive piece needs to contain before you brief it rather than after it underperforms.
Start your free NotionCue trial and map your existing coverage before planning the next quarter. Most teams find one cluster nearly complete and worth finishing, and two started and abandoned.
Before adding anything to next quarter's plan, list the topics you will explicitly not cover. A strategy without exclusions is a wish list, and the exclusions are what make the remaining capacity sufficient.
Common Questions
How many pieces should a topic cluster contain?
As many as the topic has distinct questions with independent search demand, which is usually between five and fifteen. Padding a cluster to hit a number produces thin pages that dilute it.
Should we stop publishing to focus on updating?
Rarely a full stop, but a period weighted heavily toward updating is often the right call for sites with large archives and stalled traffic. Existing pages with accumulated authority respond faster than new pages.
How do we decide between a new page and expanding an existing one?
SERP overlap again. If the new topic's results substantially overlap an existing page's target query, expand. If they do not, publish separately and link the two.