The task
You scroll through your last 30 posts and they look like 30 different blogs. One week it was a deep dive on prompt engineering, the next a vlog about your home office, then a hot take on AI hiring, then a tool review. Each post got some engagement, but your subscribers cannot finish the sentence “this person writes about ___.” Neither can you. Google has no idea what your site is about. You want 3-5 content pillars — sharp enough that every new post slots cleanly into one, and broad enough that you do not run out of things to write under any single pillar in the next 6 months.
Where AI helps — and where it does not
AI is excellent at clustering — feed it 30 titles and it will surface the latent groupings you have been drifting toward without noticing. It is also good at naming the pillars in a way that is specific (not “tech” or “productivity”). Where AI fails: deciding which pillars match your business goals versus your personal interests. The model can show you that your 30 posts cluster around “AI tools,” “founder life,” and “industry hot takes,” but it cannot tell you that hot takes drive zero subscriptions while AI tools drive 80% of them. That call is yours, with the data only you have.
A common failure mode: the model proposes pillars that are too broad — “AI,” “marketing,” “growth.” These are not pillars; they are categories. A real pillar is specific enough that you could name 3 sub-topics inside it that would not fit any other pillar. Tell the model in the prompt: tighten until the pillar fails the “could this name be any blog’s pillar” test.
What to feed the AI
- A list of your last 30 posts / videos / threads — titles only is fine, but include the rough format (essay, vlog, thread, tutorial) if formats vary
- Engagement signal per post — at least a rough rank: top 5, middle 20, bottom 5
- Your business / audience goal in one sentence — what conversion you want each post to nudge toward (subscribe, buy, refer, hire you)
- A “do not cover” list — topics you currently write about that you suspect should be retired
- Your published cadence — daily / weekly / monthly; affects how many pillars you can sustain
- The 1-2 posts you are proudest of, regardless of engagement — these often define your “voice” pillar
- The 3 reader questions you most often get in DMs or replies (these surface a pillar you might be under-writing)
- The conversion driver — which past posts have actually produced subscribes / sales
Copy-ready prompt
Cluster my past content into 3-5 sharp content pillars.
Past content (with engagement rank): {paste 30 with top 5 / middle 20 / bottom 5 marked}
Conversion driver (posts that actually produced subscribes / sales): {list which ones}
Business goal: {one sentence}
Do-not-cover list: {topics to retire}
Published cadence: {daily / weekly / monthly}
Reader questions I most often get: {list 3}
Posts I am proudest of regardless of engagement: {list 1-2}
For each pillar return:
1) Pillar name (5 words max, specific enough that it could not be any blog's pillar).
2) What it covers (1 sentence) and what it explicitly does NOT cover (1 sentence).
3) Which of my past posts belong to this pillar.
4) Sub-topics inside this pillar I have NOT written yet (5 specific post ideas).
5) Tie to business goal — how this pillar plausibly drives the conversion I named (1 sentence). If it does not, flag it as a "voice" pillar (allowed once, max).
6) Sustainability check — can I write 1 post per pillar per month for the next 6 months? If not, the pillar is too narrow.
End with:
- A "demote / archive" list — past posts that do not belong to any of the proposed pillars; recommendation per post (delete, archive, refresh-into-pillar, leave-as-orphan).
- A 6-month rough cadence — how many posts per pillar per month given my published cadence.
- The single pillar I should write the next post under, and why.Shorter variant — single-post triage
Given these content pillars: {paste pillar names + 1-line definitions}
Which pillar does this post idea belong to? — Post idea: {paste}
If it fits cleanly, write a 1-sentence justification.
If it does not fit, propose: (a) reshape the angle to fit the closest pillar, or (b) skip the post.Sample output
Good pillar names: “AI workflows for solo SaaS founders,” “Honest tool reviews (the ones I cancelled too),” “Build-in-public retros,” “Pricing math for indie products.” Specific. Each one passes the “could be any blog” test.
Bad pillar names: “Tech,” “Productivity,” “Tips,” “Marketing.” These are categories, not pillars; nothing forces a post to belong or not belong.
A useful sustainability check: “Pillar 4 (‘Pricing math for indie products’) — risky. You have 2 posts here, but pricing case studies require either real numbers or strong arguments; you can sustain 1 per month for ~4 months before you run out of fresh angles. Recommendation: combine with Pillar 1, or commit to interview-driven pricing posts (which adds prep cost).”
A useful demote list: “Demote: ‘5 productivity apps I use’ — does not belong to any pillar, weak engagement, conversion zero. Recommendation: archive. Refresh: ‘The week my deploy broke prod’ — strong personal-voice piece, fits ‘Build-in-public retros’ if you rename to make the retro angle explicit. Delete: ‘Crypto thoughts 2023’ — off-pillar, stale, drags site topical authority.”
How to refine
- If pillars are too broad: “Tighten each pillar until I can name 3 specific sub-topics inside it that would NOT fit any other pillar. If you cannot name 3, the pillar is too broad or has overlap with another.”
- If pillars miss the business goal: “Each pillar (except one allowed ‘voice’ pillar) must plausibly drive my named conversion at least 1 in 5 posts. Re-cluster if no pillar passes.”
- If too many pillars (6+): “Cut to 4. Merge the two pillars closest in audience overlap. The result will feel slightly broader; that is correct, because 6 pillars means none is the answer.”
- If pillars exclude my proudest post: “My proudest post defines the ‘voice’ pillar. Build the rest of the pillars knowing the voice pillar exists; do not force the proud post into a business pillar where it does not naturally fit.”
- If the demote list feels too aggressive: “Distinguish demote (do not write more like this), archive (keep but de-index), and refresh-into-pillar (rewrite with the pillar’s frame). Most posts should be refresh, not delete.”
Common mistakes
- Defining pillars by topic only, no goal: you produce content that ranks but does not convert; pillars without a business tie create a media business by accident.
- More than 5 pillars: you are not focused, you are sorted; a pillar count of 6+ usually means you have not made a real choice.
- Skipping the demote step: the discipline is what makes pillars work; if you let off-pillar posts stay, you teach Google and your readers that the pillars are decorative.
- One pillar with 70% of posts, the others token: the pillar map should roughly match how you’ll actually write; a 70/10/10/10 distribution means the 4-pillar structure is fake — you have one pillar and three placeholders.
- Reusing last year’s pillars when the audience shifted: pillars are descriptive of where you are going, not historical. Re-cluster quarterly.
- Forgetting the “voice” allowance: every creator has at least one off-business pillar that builds connection; calling it out as voice (not business) keeps it from competing for slots.
- No sub-topic inventory: if you cannot name 5 unwritten sub-topics inside a pillar, the pillar is too narrow to sustain or you have not thought hard enough about it.
- Equal cadence across all pillars: your strongest pillar should produce more posts than your voice pillar; equal distribution is a sign of indecision, not balance.
FAQ
- Can I have a pillar that does not drive conversion?: One out of four or five, maximum — call it your “voice” pillar. Mark it clearly so it does not eat slots from your business pillars. The voice pillar is what makes the rest of the writing feel human.
- How often should I redo this?: Quarterly. If you finish a quarter and the pillars no longer fit what you actually wrote, the pillars were wrong — not the writing. Re-cluster, do not force the writing back into stale pillars.
- What if I write across very different domains (founder + parent + photography)?: Either pick one domain for this publication and start the others on separate channels, or accept that you are running a personal brand (not a niche site) — and your pillars become the 3-4 facets of you, not the 3-4 topics of a niche. Both work; mixing them does not.
- Should pillars match my SEO strategy?: Yes, but pillars come first. Your pillars define the topic clusters; the SEO strategy fills out the long-tail under each pillar. SEO-first pillar definition produces content that ranks but does not feel like you wrote it.
- What about posts that span two pillars?: Force them into one. Cross-pillar posts dilute topical authority and confuse the reader. If the post genuinely sits between two pillars, the pillars are too close and need re-clustering.