The exact research-to-publish workflow using Claude — without writing a word manually.
⬡ What you'll build
The highest-leverage way to run a content business with one person is to use Claude for every stage of the content pipeline — from topic research to final editing — while you operate as the quality control layer, not the writer.
This lesson maps the operational workflow. Not the theory. The actual step-by-step sequence you run for every piece of content.
A content business needs, at minimum:
This workflow produces that using Claude Pro + WordPress. No additional tools. No content teams. One person, one AI subscription.
Every piece of content follows the same five stages:
1. Topic Research → What to write
2. Brief Creation → What the post covers
3. Content Generation → Draft with Claude
4. Editing Pass → Quality control
5. Publishing → WordPress + SEO basics
Each stage has a specific Claude prompt pattern and a specific human task.
Find topics that have real search demand, aren't dominated by huge sites, and match your niche.
Analyze and expand a topic list you bring in. Claude is not a keyword research tool — it cannot tell you real search volume. But it can:
After you have a list of potential topics from actual search research, bring them to Claude:
"I'm writing a content site about [niche]. Here are 10 topics I'm considering: [list]. For each one, tell me: what specific question is the searcher asking, what they already know before they search, and what answer would fully satisfy them. Then rank these from most to least specific."
Claude's output tells you which topics have a clear, answerable question (good) versus which are vague interest areas (harder to rank for).
This is the most important stage. A good brief produces good content. A vague brief produces generic content that ranks for nothing.
Topic: [exact topic]
Primary keyword: [the search phrase you're targeting]
Search intent: [what does someone searching this want to accomplish?]
Target reader: [who specifically — their knowledge level, their situation]
Competing content gaps: [what existing posts miss or get wrong]
Required sections: [specific H2s the post must cover]
Required specifics: [exact tools, versions, prices, examples that must appear]
Word count target: [minimum to fully answer the topic]
What NOT to include: [generic advice, filler sections, obvious statements]
"Help me create a content brief for a post targeting the keyword '[keyword]'. The reader is [describe reader]. Create a brief with: (1) clear search intent statement, (2) 5-7 required H2 sections that address sub-questions in depth, (3) list of specific facts, tools, prices, or examples that MUST be included for the post to be useful, (4) list of generic points this post should NOT include. Format as a structured brief I can use as input to write the post."
Review the brief. Add anything you know from personal experience. Remove anything that doesn't fit your actual niche.
With a solid brief, content generation with Claude is systematic.
"Write a [word count] word article for this brief: [paste brief].
Writing rules:
- Every section must give specific, actionable information — no filler
- Use real numbers, real tool names, real price ranges
- Write for someone who wants to actually do this, not someone who wants a summary
- No phrases like 'in today's digital landscape', 'in conclusion', 'it's important to note'
- Start each section with the most important sentence, not context-setting
- Do not hedge every statement — make clear claims
Format: Markdown with H2 and H3 headings, short paragraphs (3-4 sentences max), bullet lists only when listing genuinely discrete items."
This is critical. You must check every piece for these.
Outdated information. Claude's training data has a cutoff. Prices change. Tools get updated. Features disappear. Any specific number or product detail needs manual verification.
Fake specificity. Claude sometimes generates plausible-sounding but incorrect specific claims: "Studies show 73% of users..." — if you didn't provide that statistic, Claude may have generated it. Remove any statistic you didn't verify.
Generic phrasing in technical content. Even with strict instructions, some generic sentences slip through. Edit them out.
Wrong SEO targeting. Claude doesn't know your actual keyword strategy. It will write naturally but may optimize for the wrong phrase variation. Adjust headings and opening paragraphs to match your target keyword.
This is your main job. Not writing. Editing.
Accuracy pass:
Quality pass:
SEO pass:
Length check:
Before you publish anything, these should be configured:
/how-to-use-claude-for-seo)Running this workflow weekly:
| Day | Task | Time |
|---|---|---|
| Monday | Topic selection for the week | 30 min |
| Tuesday | Brief creation (1–2 posts) | 45 min |
| Wednesday | Generate drafts in Claude | 30 min |
| Thursday | Editing pass | 60 min |
| Friday | Publish + update internal links | 30 min |
Total: ~3–3.5 hours per week for 1–2 published posts. This is achievable alongside a full-time job.
Once the workflow is running smoothly, you can increase output without increasing time proportionally:
Batching: Create 5 briefs in one session, generate all 5 drafts back-to-back, edit them over two days.
Template refinement: As you edit, note what errors Claude makes repeatedly in your niche. Update your generation prompt to prevent them. Over time, the editing pass gets faster.
Reuse structure: Long guides can be built by generating sections separately with targeted prompts, then assembling them.
⚠The most common mistake
Using Claude output directly without an editing pass. This produces content that reads fine but fails in two specific ways: (1) outdated or invented specific claims that damage your credibility when readers notice, (2) generic quality that doesn't differentiate your site from the dozens of other AI-generated content farms covering the same topic. Your editing pass is what makes your content worth reading.
⬡Practical exercise
Run this workflow once end-to-end before optimizing anything. Pick one topic you know reasonably well, create the brief, generate the draft, do the editing pass, and publish it. The first run will take 3x longer than the rhythm above — that's expected. The second run will be faster. By the fifth run, the rhythm will feel natural.
For a content business at zero budget:
Do not spend money on stock photo subscriptions until you're generating revenue.
Implementation Checkpoint