SEO Content Strategy: Topic Clusters and Citation-Worthy Content
An SEO content strategy decides which topics your business should own, which questions each page answers, and how pages link together. Done well, it uses a small number of deep topic clusters built from real customer questions and original expertise, so both search engines and AI assistants see you as the source worth citing.
What is an SEO content strategy?
A content strategy is a decision about what you'll be known for. It answers three questions: which topics should we own, which specific questions does each page answer, and how does it all connect. Everything else (calendars, word counts, publishing cadence) is logistics.
Most small-business blogs don't have a strategy. They have a habit. Someone posts "5 tips for spring" every few weeks, nothing links to anything, and after two years there are eighty posts and four of them get traffic.
With AI search in the picture, the cost of that habit goes up. AI Overviews and assistants like ChatGPT and Perplexity favor sources that clearly know a topic in depth and say something original. Google's guidance for its AI features calls this "non-commodity content." Generic posts are, by definition, commodity content.
How we build a content strategy
Step 1: Pick three to five topics to own
Not keywords. Topics. A topic is a subject area where you have real expertise and your buyers have real questions.
For a medspa, that might be injectables, laser skin treatments and acne scarring. For a B2B software consultancy, maybe implementation, integrations and change management. For a roofer: storm damage and insurance, roof replacement, and commercial roofing.
We choose topics where three things overlap: what you sell, what your buyers search and ask AI about, and where you can say something your competitors can't.
Step 2: Mine the real questions
For each topic, we build a question list from:
- Search Console queries you already appear for
- "People also ask" and the follow-up questions AI Overviews suggest
- Keyword research tools for volume and difficulty
- Your sales calls and support inbox
- Reviews, forums and Reddit threads in your category
- What ChatGPT and Perplexity cite when asked about the topic (which shows you who you're competing with for citations)
A typical topic produces 30 to 80 distinct questions. Not all deserve a page.
Step 3: Design the cluster
Each topic becomes a cluster:
- One pillar page: the broad guide or service page. It answers the main question and links to every supporting page.
- Supporting pages: each one answers a specific question in depth. "How long does lip filler last", "Lip filler vs lip flip", "What to avoid before filler".
- Links both ways: every supporting page links back to the pillar and to two or three siblings.
This structure is how topical depth becomes visible. A crawler following your links sees thirty interconnected pages on a subject, not one lonely post.
Step 4: Audit what you already have
Before writing anything new, we sort existing content into:
- Keep: performing and accurate
- Refresh: decent topic, outdated or thin, worth improving
- Merge: two or more posts competing for the same question
- Retire: no traffic, no links, no purpose (redirected to the closest relevant page)
Refreshing and merging often produce faster wins than new content.
Step 5: Brief every page
Each page gets a brief before anyone writes a word: the primary question, the direct answer in two to four sentences, the sub-questions to cover, the original insight or example only you can provide, internal links, and the call to action. A good brief is the difference between a page that ranks and one that merely exists.
What makes content worth citing?
This is the part that separates content AI quotes from content it ignores.
- It answers first. The question as a heading, the answer immediately after. Then the nuance.
- It contains something new. Your process, your numbers, your photos, a case from last month, a mistake you see customers make. If it could have been written by someone who has never done the job, it's commodity.
- It names things. A named method ("our three-check roof assessment") is easier to reference than vague advice.
- It's structured. Comparison tables, numbered steps, short paragraphs. Easier for humans to scan and for machines to extract.
- It has a real author. A byline with a person's name, role and credentials, and an author page.
- It's current. Dates, prices and regulations get checked and updated, with a visible "updated" date.
- It's honest about trade-offs. "This isn't right for you if..." builds trust with readers and reads as expertise.
What it doesn't need
Following Google's guidance, you don't need to chop content into tiny chunks for AI, write separate "AI versions" of pages, or create a page for every long-tail variation of a keyword. One strong page that answers a question thoroughly beats five thin variants.
Using AI in your content process
Plenty of teams use AI tools for research, outlines and first drafts, and that's fine. Google's guidance focuses on whether content is helpful, not how it was produced. (source) The practical risk is that AI drafts gravitate to the average answer. If your page says the same thing as every other page, there's no reason to rank or cite it.
If you want help setting up a sensible AI-assisted content workflow for your own team, with brand-voice guidelines and review steps, that's part of our marketing automation work.
What a content plan looks like in practice
A realistic plan for a small business:
- Three to five topic clusters, each with a pillar and a first set of supporting pages
- A refresh list for existing content, prioritized by opportunity
- A publishing rhythm you can actually sustain (consistent beats ambitious)
- A brief for every page before it's written
- Quarterly review: what's ranking, what's being cited, what to expand
How do you measure content performance?
- Organic traffic and leads by cluster, not just by post
- Rankings and AI Overview appearances for the questions each page targets
- Citations in ChatGPT and Perplexity for cluster questions, from our GEO tracking
- Assisted conversions: content that readers saw before contacting you
- Engagement signals like scroll depth and onward clicks
Proof it works
S Bertram Foods had no content program at all, just word of mouth. Keyword research, optimized product pages and a content calendar of genuinely useful articles helped take them to 135,000 sessions and a 235% increase in organic traffic over 12 months. Read the S Bertram Foods case study.
Content production is part of our SEO, AEO and GEO engagement, not a separate upsell. If you want to know which topics you could realistically own, start with an SEO audit.
Frequently asked questions
How many blog posts do I need?
Fewer than you think. A dozen well-connected, genuinely useful pieces on a topic you want to own usually beat sixty thin posts on everything. Volume without depth mostly creates pages nobody visits.
What is a topic cluster?
A pillar page covering a broad topic, supported by detailed pages that each answer one specific question, all linked to each other. It makes your expertise on the topic obvious to readers and crawlers.
Can we use AI to write our content?
You can use AI to help, and many teams do. Google cares about whether content is helpful and original, not how it was typed. The risk is sameness: AI drafts tend to say what everyone else says, and that's exactly the content AI search has no reason to cite. Your expertise has to be in it.
Who writes the content?
Your choice. We can write it in your voice after interviewing your experts, edit what your team drafts, or give your writers briefs. Most clients prefer we write and they review for accuracy.
Should we update old posts or write new ones?
Usually update first. Old posts that already have some rankings and links can often be improved faster than new pages can be built from zero. We audit what you have before planning anything new.