# SEO Content Strategy Workflow — Instructions for Claude ## How to use this file **Starting for the first time:** ``` Read @CONTENT_STRATEGY_WORKFLOW.md and let's begin. ``` Claude will read the file, collect your company details, and start Stage 1. **Resuming after ending a session:** ``` Read @CONTENT_STRATEGY_WORKFLOW.md and @WORKFLOW_LOG.md and let's pick up where we left off. ``` **Before ending any session, run this to save your state:** ``` Update WORKFLOW_LOG.md: where we are in the workflow, every decision made this session, what's been approved, and exactly what to do when I start a new session. ``` Commit `WORKFLOW_LOG.md` to git regularly so your state is never lost. --- ## Your role You are working through a structured 6-stage SEO content strategy workflow. Read this entire file before starting. Your job is to guide the user through each stage in sequence, writing all outputs to files as specified. --- ## Setup: Before you start Ask the user to confirm the following. Replace every [BRACKET] in this file with their answers before running any prompts. | Variable | Description | |---|---| | [COMPANY NAME] | The company you're building content for | | [COMPANY DESCRIPTION] | One sentence: what they do and who they serve | | [PILLAR COUNT] | Number of content pillars (default: 3–5) | | [CADENCE] | Publishing cadence (default: 1 pillar + 3 clusters per week) | | [START DATE] | When the first article publishes | Keep a running log at `WORKFLOW_LOG.md`. Record every significant decision — approved pillars, keyword choices, strategic calls, what stage you're at — so any new session can pick up context without reconstructing it. At the end of each stage, summarise what was produced and wait for explicit approval before moving to the next stage. --- ## Stage 1: Discovery & Strategy **Run in:** Claude Chat (Opus preferred) **Outputs:** `STRATEGY.md`, `VOICE_GUIDE.md` --- ### Prompt 1A — Initial Discovery You are a senior content strategist specialising in B2B/B2C tech marketing. Your job is to help me build a content strategy that drives SEO traffic and answers the questions my target audience is actually asking. Before we build anything, I need you to interview me. Ask me questions one section at a time — wait for my answer before moving to the next section. Cover: 1. The company and product — what we do, who we serve, what problem we solve 2. Our target audience — who they are, their job titles, their pain points, what they search for 3. Our competitive landscape — who else is creating content in this space, what they do well, where the gaps are 4. Our current content situation — what we already have, what's working, what's not 5. Our goals — what does success look like? Traffic? Leads? Brand authority? After I've answered all sections, synthesise what I've told you and reflect it back to me as a "strategic foundation" summary before we move to building the content pillars. Push back on anything that seems unclear or contradictory. **Checkpoint:** Present the strategic foundation. Wait for confirmation before running Prompt 1B. --- ### Prompt 1B — Content Pillars Based on everything we've discussed, I want you to propose a content pillar strategy. A content pillar is a broad topic we can own — it should be: - Directly relevant to what [COMPANY NAME] does - Something our target audience actively searches for - Defensible — not so broad that we're competing with Wikipedia, not so narrow we run out of things to say - Connected to our product or service without being purely promotional Propose 3–5 content pillars. For each one, give me: - The pillar name and a one-sentence description - Why this topic is strategically valuable for us - The type of reader this pillar attracts and where they are in the buying journey - 3 example cluster topics that would sit under it - Any risks or weaknesses (e.g. high competition, commoditised content) Then give me your recommendation on which pillar to prioritise first and why. Be direct. If a pillar idea is weak, say so. **Checkpoint:** Present the pillars. Wait for approval before running Prompt 1C. --- ### Prompt 1C — Voice & Style Guide I need to define the editorial voice and style for all content we produce. This will become a reference document every piece of content is checked against. Based on what you know about [COMPANY NAME] and our audience, draft a voice and style guide. Include: 1. Voice — 3–4 adjectives that describe our tone, each with a brief explanation and a "we sound like this / we don't sound like this" example 2. Audience respect — how technical can we be? What do we assume they already know? 3. Things we always do — e.g. use active voice, lead with the reader's problem, use real examples 4. Things we never do — e.g. use jargon without explanation, make claims we can't back up, write long intros that bury the point 5. A note on promotional balance — how do we mention our product without every post feeling like an ad? Format this so it can be dropped directly into a VOICE_GUIDE.md file. **After this prompt:** Write the strategic foundation from Prompt 1A to `STRATEGY.md`. Write the voice guide to `VOICE_GUIDE.md`. Log completion of Stage 1 in `WORKFLOW_LOG.md`. **Checkpoint:** Confirm both files are written. Wait for approval before moving to Stage 2. --- ## Stage 2: SEO & Keyword Research **Run in:** Claude Code (Sonnet, with web search enabled) **Bring:** `STRATEGY.md` and `VOICE_GUIDE.md` from Stage 1 **Outputs:** `/pillars/[name]/keywords.md` per pillar, `SEO_PLAN.md` --- ### First: write CLAUDE.md Write this file to the repo root. Claude Code reads it at the start of every session and it keeps every session calibrated without re-pasting instructions. ``` # CLAUDE.md ## About this project This repository contains the content strategy and all articles for [COMPANY NAME]'s blog. All content is written in markdown. ## Your role You are a senior content strategist and SEO specialist. Every session, read STRATEGY.md and VOICE_GUIDE.md before doing anything else. ## Content standards - All content must follow the voice and style rules in VOICE_GUIDE.md - All articles must target a primary keyword defined in their keywords.md file - Internal links should follow INTERLINKING_PLAN.md once it exists - Never write promotional content — our product can be mentioned but never pushed ## File conventions - Drafts go in draft.md - Approved final versions go in final.md - Never overwrite final.md without being explicitly asked to ## SEO rules - Primary keyword in H1, first paragraph, and at least two H2s - Meta description: 150–160 characters, includes primary keyword - Aim for 1 internal link per 300 words once INTERLINKING_PLAN.md exists ``` --- ### Prompt 2A — Pillar-Level Keyword Research Run once per pillar. Replace [PILLAR NAME] each time. Read STRATEGY.md and identify the pillar called [PILLAR NAME]. I want you to do keyword research for this pillar. Your goal is to find: 1. A primary keyword for the pillar itself — the broadest, highest-intent term that a pillar article could target 2. 10–15 related keywords and phrases that cluster around this topic — these will inform cluster article topics 3. For each keyword: your assessment of search intent (informational, navigational, commercial, transactional), competition level (low/medium/high — reason this out from what you find), and who is likely searching for it Use web search to: - Look at what content currently ranks for the main topic - Identify what questions people are asking (look for "people also ask" patterns, forums, Reddit, community sites) - Note which competitors from STRATEGY.md are ranking and for what [OPTIONAL — if you have SEO tool data: I'm pasting in keyword data from [Ahrefs/SEMrush] below. Use this to validate or override your estimates on search volume and competition. {PASTE DATA HERE}] When you're done, write your findings to /pillars/[pillar-name]/keywords.md using this structure: ``` # Keywords: [Pillar Name] ## Primary keyword [keyword] — [intent] — [competition assessment] — [who's searching] ## Keyword clusters | Keyword | Intent | Competition | Notes | |---|---|---|---| ## Competitor content notes [What's ranking, who owns it, where the gaps are] ## Opportunities [Your top 3 observations about where we can win] ``` Repeat this process for all [PILLAR COUNT] pillars in STRATEGY.md, writing a separate keywords.md for each one. --- ### Prompt 2B — SEO Plan Overview Run this after all pillar keyword files exist. Read all the keywords.md files across /pillars/. I want you to produce a master SEO_PLAN.md that synthesises the keyword research into a coherent SEO strategy. Include: 1. **Keyword priority matrix** — rank all primary keywords by opportunity (a combination of relevance, competition, and our realistic ability to rank given we're a new blog with no existing authority) 2. **Search intent map** — group our target keywords by intent stage. Which keywords attract people early in their research? Which attract people close to a decision? 3. **Competitive gaps** — where are competitors weak or absent that we can move into quickly? 4. **Quick wins vs long-term plays** — which keywords could we realistically rank for in 3–6 months vs which are 12+ month bets? 5. **Risks** — any areas where the content landscape is too crowded, or where we'd be writing into a black hole Write this to SEO_PLAN.md in the repo root. Be direct — if a pillar has weak keyword opportunity, say so and suggest what to do about it. --- ### Prompt 2C — Cluster Topic Identification Run this after SEO_PLAN.md exists. Read STRATEGY.md, SEO_PLAN.md, and all /pillars/[name]/keywords.md files. For each pillar, I want you to define the cluster articles that will sit underneath it. Each cluster article should: - Target one specific keyword or question from the keyword clusters we identified - Serve a specific reader need (not just stuff keywords) - Be distinct enough from other clusters that there's no cannibalisation - Connect logically back to the pillar topic For each pillar, propose 4–6 cluster article topics. For each cluster, give me: - Proposed article title (written as a reader would search for it) - Primary keyword it targets - One sentence on what the article covers and why someone would read it - Where in the buying journey this reader sits Present this as a structured list I can review before we build briefs. Do not write to file yet — I want to approve the cluster list first. **Checkpoint:** Present the cluster topics. Wait for approval before moving to Stage 3. Log approved clusters in `WORKFLOW_LOG.md`. --- ## Stage 3: Content Map, Briefs & Posting Schedule **Run in:** Claude Code (Sonnet) **Bring:** Approved cluster topic list, `SEO_PLAN.md`, all `keywords.md` files, `STRATEGY.md` **Outputs:** `CONTENT_MAP.md`, all `brief.md` files, `POSTING_SCHEDULE.md`, `briefs-audit.md` --- ### Prompt 3A — Build the Content Map Read STRATEGY.md, SEO_PLAN.md, and all keywords.md files in /pillars/. I have approved the cluster topics. Build a CONTENT_MAP.md that maps the full content architecture. Structure it as follows: For each pillar: - Pillar name, primary keyword, and one-line description - List of cluster articles sitting under it, each with their primary keyword - A note on the logical reading journey — how does someone move from a cluster article to the pillar and back? Then add a section called "Content Relationships" that notes: - Which clusters serve as entry points (high search volume, top of funnel) - Which clusters are conversion-adjacent (reader is close to a decision) - Any clusters that bridge two pillars Format this so it's easy to scan — use tables where it helps. Write to CONTENT_MAP.md in the repo root. **Checkpoint:** Present the content map. Wait for approval before running Prompt 3B. --- ### Prompt 3B — Generate All Briefs Read CONTENT_MAP.md, SEO_PLAN.md, STRATEGY.md, and VOICE_GUIDE.md. I want you to generate a brief for every piece of content in the content map — every pillar article and every cluster article. For each piece, spawn a sub-agent that: 1. Reads the relevant keywords.md file for that pillar 2. Writes a complete brief to the correct file path Pillar briefs go to: /pillars/[pillar-name]/brief.md Cluster briefs go to: /clusters/[cluster-name]/brief.md Every brief must follow this exact structure: ``` # Brief: [Article Title] ## Overview - **Type:** Pillar / Cluster - **Pillar:** [which pillar this belongs to] - **Primary keyword:** [exact keyword] - **Secondary keywords:** [2–3 supporting terms] - **Target reader:** [specific description — job title, situation, what they're trying to figure out] - **Search intent:** [what is this person actually trying to do] - **Funnel stage:** Top / Middle / Bottom - **Estimated word count:** [range — e.g. 1800–2200 for clusters, 3000–4000 for pillars] ## The reader's problem [2–3 sentences from the reader's perspective, not the brand's.] ## What this article must do [3–5 bullet points — what does the reader know or believe by the end?] ## Recommended structure [Proposed H1, H2s, and H3s — written as a reader would see them] ## Key points to cover [The substance of the article — arguments, explanations, examples, data] ## What to avoid [Anything that would make this article generic, off-brand, or cannibalise another piece] ## Internal linking opportunities [TBD — will be populated in Stage 4] ## SEO notes - Meta description: [150–160 characters including primary keyword] - H1 suggestion: [strong H1 with primary keyword naturally included] - Any specific SEO considerations for this topic ``` After all briefs are written, report back with a summary: how many briefs were created, any gaps or conflicts you noticed, any briefs where the keyword opportunity seemed weak. --- ### Prompt 3C — Posting Schedule Read CONTENT_MAP.md and all brief.md files across /pillars/ and /clusters/. I want you to build a posting schedule. Here are my constraints: - Publishing cadence: [CADENCE] - Start date: [START DATE] - Total pieces to schedule: [NUMBER — count from the content map] - [OPTIONAL: any pieces that must go live by a specific date] When sequencing, apply this logic: 1. Clusters that are entry points (high volume, top of funnel) should publish early — they drive traffic that validates the strategy 2. Pillar articles benefit from having at least 2–3 clusters live first — so internal links exist when the pillar launches 3. Conversion-adjacent clusters should publish after the pillar they sit under 4. Spread pillars evenly — don't cluster all content from one pillar in consecutive weeks Build the schedule as a table: | Week | Publish date | Content type | Article title | Pillar | Primary keyword | Notes | Add a rationale section explaining the sequencing logic and any trade-offs you made. Write to POSTING_SCHEDULE.md in the repo root. --- ### Prompt 3D — Brief Quality Audit Run this after all briefs are written, before moving to Stage 4. Read all brief.md files across /pillars/ and /clusters/, plus STRATEGY.md and SEO_PLAN.md. Audit every brief against these criteria and give me a report: 1. **Keyword conflicts** — are any two briefs targeting the same or near-identical primary keyword? Flag these as cannibalisation risks. 2. **Coverage gaps** — are there topics in SEO_PLAN.md that no brief addresses? 3. **Brief quality** — for each brief, flag if any of these are true: - The reader problem is vague or brand-centric rather than audience-centric - The recommended structure is too generic (could apply to any article on this topic) - The word count estimate seems wrong for the complexity of the topic - The funnel stage seems misclassified 4. **Strategic coherence** — does the full set of briefs tell a coherent story about what [COMPANY NAME] knows and cares about? Be specific. Name the brief, name the problem, suggest the fix. Write your findings to /briefs-audit.md. **Checkpoint:** Review the audit with the user. Resolve critical issues before moving to Stage 4. --- ## Stage 4: Interlinking Plan **Run in:** Claude Code (Sonnet) **Bring:** All completed briefs, `CONTENT_MAP.md`, `POSTING_SCHEDULE.md` **Outputs:** `INTERLINKING_PLAN.md`, updated briefs --- ### Prompt 4A — Map the Link Architecture Read CONTENT_MAP.md, POSTING_SCHEDULE.md, and all brief.md files across /pillars/ and /clusters/. I want you to define the interlinking architecture for our content set. Work in two layers: **Layer 1: Structural links** — mandatory, appear in every article. For each pillar article: - It must link down to every cluster article in its pillar - List the recommended anchor text for each cluster link (natural language — not "click here", not keyword-stuffed) For each cluster article: - It must link up to its parent pillar - It must link to at least one sibling cluster in the same pillar - List recommended anchor text for each **Layer 2: Contextual link opportunities** — opportunistic, where topics naturally overlap. For each contextual opportunity, note: - Source article (the one containing the link) - Target article (the one being linked to) - The likely context — what sentence or paragraph would this link appear in - Suggested anchor text Present Layer 1 as a table and Layer 2 as a structured list grouped by source article. Do not write to file yet — present this for review first. **Checkpoint:** Review with the user. Once approved, run Prompt 4B. --- ### Prompt 4B — Write the Interlinking Plan to File The interlinking architecture looks good. Write it to INTERLINKING_PLAN.md in the repo root. Structure the file so it can be used in two ways: 1. **By article** — a writer looking up any given article sees exactly which links to include, where, and what anchor text to use 2. **As a master map** — a table showing every link relationship in the content set at a glance Add an introduction section explaining the interlinking logic to anyone picking up this document cold. Also add a section called "Links to add retrospectively" — contextual links that reference articles not yet published. Flag them so writers know to add these links once the target article goes live. Write to INTERLINKING_PLAN.md. --- ### Prompt 4C — Update All Briefs with Interlinking Sections Read INTERLINKING_PLAN.md and all brief.md files. For each brief, update the "Internal linking opportunities" section with the specific links that article should contain — both structural and contextual — including the recommended anchor text. Make only this change to each brief. Do not alter any other section. After updating all briefs, report back with a count of how many briefs were updated and flag any article that ended up with fewer than 2 internal links. **Checkpoint:** Confirm all briefs are updated. Log completion of Stage 4 in `WORKFLOW_LOG.md`. Move to Stage 5. --- ## Stage 5: Writing **Run in:** Claude Code (Sonnet for clusters, Opus for pillars) **Bring:** Everything in the repo — briefs, keywords, interlinking plan, voice guide, CLAUDE.md **Outputs:** `draft.md` and `final.md` per article Model guidance: - **Pillar articles:** use Opus. These are flagship pieces, 3,000–4,000 words. Worth the higher cost. - **Cluster articles:** use Sonnet. 1,500–2,500 words, faster and significantly cheaper. - **On usage caps:** if you're on a subscription rather than pay-per-token, batch cluster drafts across sessions rather than running all in parallel. Parallel sub-agents drain allowances fast. --- ### Prompt 5A — Draft a Pillar Article Read the following files before writing anything: - CLAUDE.md - VOICE_GUIDE.md - /pillars/[PILLAR-NAME]/brief.md - /pillars/[PILLAR-NAME]/keywords.md - INTERLINKING_PLAN.md You are writing the pillar article for [PILLAR NAME]. This is a flagship piece — it is the most comprehensive treatment of this topic we will publish. It needs to be genuinely useful to an expert reader, not just a surface-level overview. Before you draft, tell me: 1. Your interpretation of what the reader's core problem is 2. The argument or throughline of the article — what is the intellectual spine that holds it together? 3. Any gaps you notice in the brief that we should resolve before writing Wait for my confirmation before drafting. When confirmed, write the full article following the structure in the brief. Apply these rules: - Open with the reader's problem, not with context-setting or company positioning - Every H2 should advance an argument, not just introduce a new subtopic - Use concrete examples — invented but plausible is fine, clearly labelled - Include all internal links from the brief's interlinking section, with the anchor text specified - End with a clear "what to do next" — not a sales pitch, a genuine next step for the reader - Write the meta description as specified in the brief Write to /pillars/[PILLAR-NAME]/draft.md After writing, give me a brief self-assessment: what you think works well, what feels weakest, and what you'd want a human editor to pay most attention to. --- ### Prompt 5B — Draft a Single Cluster Article Read the following files before writing anything: - CLAUDE.md - VOICE_GUIDE.md - /clusters/[CLUSTER-NAME]/brief.md - /clusters/[CLUSTER-NAME]/keywords.md - /pillars/[PARENT-PILLAR-NAME]/brief.md - INTERLINKING_PLAN.md Write the cluster article for [CLUSTER NAME]. This article sits under the [PARENT PILLAR] pillar. The reader finding this article is at [FUNNEL STAGE — copy from brief]. Apply these rules: - The opening paragraph must contain the primary keyword naturally and establish the reader's problem within the first 3 sentences - Each H2 covers one distinct idea — no H2 should be trying to do two things - Be specific: every recommendation needs a concrete explanation of how - Include all internal links from the brief's interlinking section with the specified anchor text - Close with one clear takeaway — the single most important thing the reader should remember or do - Write the meta description as specified in the brief Write to /clusters/[CLUSTER-NAME]/draft.md --- ### Prompt 5C — Batch Cluster Drafting (same pillar) Use this to draft all clusters for one pillar in a single pass. Spawns sub-agents so contexts stay clean. **Path note:** Any time a sub-agent is told to "read its own brief.md," the path must be explicit — sub-agents start cold and don't know the repo structure. The pattern is /clusters/[cluster-slug]/brief.md. List the actual slugs below. Read the following files: - CLAUDE.md - VOICE_GUIDE.md - INTERLINKING_PLAN.md - /pillars/[PILLAR-NAME]/brief.md I want to draft all cluster articles for the [PILLAR NAME] pillar in one pass. The clusters are: [LIST CLUSTER NAMES AND THEIR FULL FOLDER PATHS — e.g. /clusters/async-code-review/] Spawn a sub-agent for each cluster. Each sub-agent should: 1. Read its own brief at /clusters/[cluster-slug]/brief.md and keywords at /clusters/[cluster-slug]/keywords.md 2. Draft the full article following the rules below 3. Write to /clusters/[cluster-slug]/draft.md 4. Report back a one-paragraph summary of what it wrote Rules every sub-agent must follow: - Follow VOICE_GUIDE.md - Open with the reader's problem within the first 3 sentences - Include all internal links from the interlinking section of its brief - Be specific — no generic advice - Write the meta description from the brief - Do not mention [COMPANY NAME] more than twice unless the brief specifically calls for it When all sub-agents are done, give me a summary table: | Cluster | Word count | Primary keyword used in H1? | Internal links included? | Any issues? | --- ### Prompt 5D — SEO Pass Read the following: - /[pillars or clusters]/[ARTICLE-NAME]/draft.md - /[pillars or clusters]/[ARTICLE-NAME]/keywords.md - /[pillars or clusters]/[ARTICLE-NAME]/brief.md - CLAUDE.md Run an SEO audit on this draft and make corrections directly in the file. Check and fix: 1. **Primary keyword** — appears in H1, within the first 100 words, in at least two H2s, and in the meta description. If missing, add it naturally — do not keyword-stuff. 2. **Secondary keywords** — each appears at least once in the body. If missing, find a natural place. 3. **Meta description** — 150–160 characters, includes primary keyword, written to earn a click. Rewrite if needed. 4. **H1** — contains primary keyword, written as a compelling headline. Rewrite if it reads like an internal label. 5. **Internal links** — every link specified in the brief's interlinking section is present with the correct anchor text. Add any that are missing. 6. **Readability** — flag any paragraph longer than 5 sentences. Flag any sentence longer than 35 words. Do not rewrite these — flag them for the human editor. After making changes, write a brief changelog: what you changed and why. Save the updated version back to draft.md. --- ### Prompt 5E — Move to Final The draft at /[pillars or clusters]/[ARTICLE-NAME]/draft.md has been approved. Do the following: 1. Copy the content to /[pillars or clusters]/[ARTICLE-NAME]/final.md 2. Do a final check — confirm the meta description is present at the top of the file, all H tags are correctly formatted in markdown, and all internal links are formatted correctly as markdown links 3. Report any formatting issues you found and corrected Do not make any content edits — only formatting and structural checks. --- ## Stage 6: Critical Review **Run in:** Claude Chat (Opus) **Purpose:** Harsh editorial reads that surface what's actually wrong, not what could theoretically be better. Paste content directly into Chat rather than reading from files — it forces you to look at the article the way a reader does, as a standalone piece of text, not a file in a repo. --- ### Prompt 6A — Strategy Review Run this after Stage 1, before committing to the full content map. I'm going to share my content strategy with you. I want you to read it as a skeptic, not a collaborator. Your job is not to help me feel good about this strategy. Your job is to find the ways it will fail. After reading, tell me: 1. **The central weakness** — if this strategy has one fatal flaw, what is it? Be specific. "The pillars are too broad" is not specific. "Pillar 2 targets a keyword space dominated by [competitor], and nothing in this strategy explains how a [company type] competes with that" is specific. 2. **Differentiation problem** — would a reader be able to tell this content came from [COMPANY NAME] specifically, or could any company in this space have written it? 3. **Audience mismatch risks** — where are we writing for who we think our customer is rather than who they actually are? 4. **What's missing** — what topic, question, or content type is conspicuously absent that our audience almost certainly cares about? 5. **The hardest question** — what is the one question about this strategy I'm probably hoping you won't ask? Ask it. Do not soften this with praise. Do not tell me what's working unless I ask. I want the problems. Here is the strategy: [PASTE STRATEGY.md CONTENT] --- ### Prompt 6B — Pillar Article Review Read this as if you are: a [TARGET JOB TITLE] who found this article through a Google search for "[PRIMARY KEYWORD]". You did not seek out this company. You have no loyalty to them. You are busy and skeptical. You will stop reading the moment the article stops being useful. After reading, give me: 1. **The verdict** — one paragraph. Would you finish reading? Would you remember anything from it tomorrow? Would you share it or come back to this site? Be honest. 2. **Where you would have stopped reading** — identify the exact moment the article loses a reader like you. Quote the line or section. Explain why. 3. **The generic problem** — identify every place where the advice, explanation, or example could have appeared in any article on this topic. List them. 4. **Claims that need proof** — identify every assertion a skeptical reader would push back on. List every claim that isn't backed up. 5. **The structure problem** — does the article build toward something, or is it just a list of related points? If the H2s could be rearranged in any order without losing meaning, the article has no spine. Tell me if that's true here. 6. **The intro** — read only the first paragraph. Does it earn the second paragraph? Rewrite it if it doesn't. 7. **The ending** — does it end or does it just stop? What should the reader do or think after finishing? Do not tell me what works. Tell me what doesn't. Here is the article: [PASTE DRAFT CONTENT] Here is the brief it was written against: [PASTE BRIEF CONTENT] --- ### Prompt 6C — Cluster Article Review Read this cluster article as a critical editor. The target reader is [TARGET READER FROM BRIEF] searching for "[PRIMARY KEYWORD]". I want four things from you: 1. **Does it answer the question?** By the end, does the reader have a complete, usable answer? If not, what's missing? 2. **The specificity test** — for every piece of advice or recommendation: could the reader actually do this based on what's written, or would they need to go elsewhere? List every piece of advice that fails this test. 3. **Cannibalisation check** — does this article stray into territory that belongs to another article? Flag any section that overlaps. 4. **The one cut** — if you had to cut 20% of this article to make the remaining 80% stronger, what would you cut and why? Here is the article: [PASTE DRAFT CONTENT] Here is the content map for cannibalisation reference: [PASTE CONTENT_MAP.md] --- ### Prompt 6D — Full Content Set Review Run this once you have 4–6 articles in final draft. I'm going to share a set of articles. Read them as a body of work, not individually. Tell me: 1. **Voice consistency** — does this read like one publication with a clear point of view, or does it feel like different writers with different agendas? Where does the voice drift most noticeably? 2. **Depth consistency** — are all articles pulling their weight equally? Name the weakest piece and explain why. 3. **The gap a competitor would exploit** — if a competitor read this entire content set, what obvious adjacent topic would they write to steal our readers? 4. **Reader journey** — pick a reader who lands on [CLUSTER ARTICLE NAME] from Google. Walk me through where the internal links would take them. Does that journey make sense? 5. **What this content set says about the company** — based only on this content, what does a reader conclude about what [COMPANY NAME] believes, knows, and cares about? Is that the impression you want? Here are the articles: [PASTE EACH FINAL ARTICLE WITH ITS TITLE AS A HEADER] --- ### Prompt 6E — Self-Aware Revision Run this after receiving any critical review. I've just received this critique of my article: [PASTE CRITIQUE] Here is the current draft: [PASTE DRAFT] I want you to do two things: First — for each problem identified in the critique, tell me whether you agree it's a real problem and why. If you disagree with any criticism, say so and make the case. Don't just accept the critique because it was given. Second — for the problems you agree are real, propose the specific fix. Not "strengthen this section" — show me the rewritten version. For structural problems, show me the revised outline. For weak openings, write me three alternative opening paragraphs to choose from. Prioritise the fixes in order of impact. --- ## Weekly rhythm (once the system is running) 1. Check `POSTING_SCHEDULE.md` — what's due this week 2. Run Prompt 5B or 5C for that week's clusters (Sonnet) 3. Run Prompt 5A for the pillar if one is due (Opus) 4. Run Prompt 5D — SEO pass on all drafts 5. Human reads and edits 6. Run Prompt 5E — move approved pieces to final.md 7. Publish Stage 6 (critical review) runs alongside this loop as a quality gate before anything moves to final. --- ## Repo structure reference ``` / ├── CLAUDE.md ← style guide and SEO rules, read by Code every session ├── CONTENT_STRATEGY_WORKFLOW.md ← this file ├── WORKFLOW_LOG.md ← running log of decisions and current stage ├── STRATEGY.md ← pillars, audience, positioning ├── VOICE_GUIDE.md ← tone and brand voice ├── SEO_PLAN.md ← keyword strategy overview ├── CONTENT_MAP.md ← full content hierarchy ├── INTERLINKING_PLAN.md ← link architecture ├── POSTING_SCHEDULE.md ← publishing calendar │ ├── /pillars │ └── /[pillar-name] │ ├── brief.md │ ├── keywords.md │ ├── draft.md │ └── final.md │ └── /clusters └── /[cluster-name] ├── brief.md ├── keywords.md ├── draft.md └── final.md ```