SEO used to require a team: a content writer, an SEO specialist, a technical auditor, someone managing the keyword spreadsheet. With Claude Code, one person can handle the entire pipeline — from keyword research to published, optimised blog post — in a fraction of the time.
This isn't theoretical. We use this exact workflow to produce the blog you're reading right now. Here's how it works.
The workflow at a glance
- Keyword research — find low-competition keywords with real search volume
- Keyword clustering — group related terms so one post ranks for dozens of queries
- Competitor analysis — study what's already ranking and reverse-engineer the formula
- Content creation — write the post with personality, not AI slop
- On-page SEO — apply the checklist that search engines reward
- Technical SEO — make sure the infrastructure doesn't undermine your content
Claude Code can help with every step. Let's break them down.
Step 1: Keyword research
You still need a dedicated SEO tool for this — SEMrush, Ahrefs, or similar. AI can't replace real search volume data. What Claude Code excels at is processing the output.
Export your keyword list as a CSV, drop it in your project, and Claude can:
- Filter for keywords with difficulty under 30 and volume over 100
- Separate informational keywords (blog posts) from commercial ones (landing pages)
- Flag keywords that aren't relevant to your business
- Suggest which keywords to prioritise based on competition and intent
The rule of thumb: keyword difficulty 30 or below means you have a realistic shot at ranking. Below 10 means you could reach the top three relatively quickly.
Step 2: Keyword clustering
This is where most people leave money on the table. A single blog post shouldn't target just one keyword — it should target a cluster of related terms.
Give Claude your target keyword and ask it to build a cluster. For example, starting with "clogged drain fix," a cluster might include: "unclog kitchen sink," "slow drain bathroom," "home remedy for clogged drain," "what dissolves hair in drains."
One well-written post can rank for 50 or more keywords. That's 50 chances for someone to find you through search.
Step 3: Competitor analysis
Before writing anything, look at what's already winning. Search your target keyword, find the top three ranking blog posts (skip Reddit and forums), and study them.
Claude Code can help you extract the patterns:
Look at these three URLs and tell me:
- Average word count
- Number of H2 headings
- Key topics each one covers
- What they have in common structurally
This gives you a template. Not to copy, but to understand the format that Google currently rewards for this keyword. Then beat it — better structure, more depth, more personality.
Step 4: Content creation (the hard part)
This is where 90% of AI-generated content fails. The default output from any AI is what the SEO community calls "AI slop" — generic, personality-free text that sounds like it was written by a committee of beige cardigans.
The fix: give Claude reference material for your voice.
Create files in your project for:
- Voice examples — samples of writing you've done (emails, LinkedIn posts, anything)
- Stories — real anecdotes from your work
- Opinions — your actual takes on industry topics
- Humor style — examples of tone that matches your brand
Then tell Claude to use them. The difference is dramatic. Instead of "maintaining your plumbing system is essential for homeowner peace of mind," you get something a human would actually want to read.
If you want a deep dive on keeping AI output sounding human, our Anti-Slop Checker catches AI-written patterns and rewrites them.
Step 5: On-page SEO
There's a checklist of 80+ signals that search engines look for. The most important ones:
- Primary keyword in the first 100 words of the post
- Exactly one H1 tag — your title
- H2 and H3 hierarchy — structured headings, not random bold text
- 3-5 internal links — linking to other pages on your site
- 2-3 external links — linking to relevant, authoritative external sources
- 4-8 FAQ questions — these can capture featured snippets
- Meta title and description — what appears in Google search results
- Image alt text — descriptive, keyword-relevant where natural
Claude Code can apply this entire checklist to a finished post. The key instruction: "Optimise for these SEO signals while maintaining the voice and personality of the original post." Without that second half, it'll strip out everything interesting and produce search-engine-friendly cardboard.
If you want the checklist enforced deterministically rather than relying on Claude remembering, a PostToolUse hook can run it as a gate after every post edit — same pattern as auto-formatting code, applied to SEO.
Step 6: Technical SEO
Your content can be perfect, but if the technical foundation is broken, Google can't find or index it. The essentials:
Static Site Generation
This is non-negotiable. Your pages need to be pre-rendered so Google gets instant HTML, not a loading spinner. Next.js with generateStaticParams() handles this automatically for blog posts.
Sitemap
Without sitemap.xml, Google doesn't know your pages exist. Generate it automatically — Claude Code can set this up in minutes.
Lighthouse scores
Google Lighthouse scores your site on Performance, Accessibility, Best Practices, and SEO. Aim for 100 across all four. Run the audit, paste the report into Claude Code, and ask it to fix every issue. Usually takes one or two iterations.
Meta tags
Every page needs proper Open Graph and Twitter Card meta tags. These control how your content appears when shared on social media — and they signal professionalism to search engines.
Turning this into a repeatable system
The real power isn't doing this once. It's turning the whole workflow into a system that you can repeat daily.
Once you've refined your process — keyword selection, voice references, on-page checklist, technical setup — package it into a Claude Code skill. A skill encodes the entire workflow so you can produce an optimised blog post with a single conversation. The skills marketplace guide covers where these live — Anthropic's examples, community lists, and curated libraries — and how to install one wherever you found it.
The publishing cadence matters too. Don't dump 50 posts at once — Google flags sudden volume spikes. Start with one post per day, gradually increase to two or three over a few weeks.
What this looks like in practice
For this site (Skill Locker), we:
- Research keywords around Claude Code, AI coding, and productivity
- Build keyword clusters for each target keyword
- Write posts that genuinely help developers — not keyword-stuffed filler
- Apply the on-page SEO checklist to every post
- Use static site generation so Google indexes instantly
- Interlink posts to build topical authority
The posts you're reading right now are the result of this exact workflow.
FAQ
Does this work for any industry?
Yes. The workflow is the same whether you're writing about plumbing, SaaS, fitness coaching, or AI coding tools. The keywords change; the process doesn't.
Do I need SEMrush specifically?
No. Ahrefs, Ubersuggest, and other keyword research tools work. You need real search volume data — that's the part AI can't generate reliably on its own.
How long before I see results?
SEO is a long game. Expect to wait 2-4 weeks for pages to get indexed, and 2-6 months for meaningful ranking improvements. The compound effect is what matters — consistent publishing builds authority over time.
Will Google penalise AI-generated content?
Google penalises low-quality content, regardless of how it's produced. AI-generated slop gets penalised. AI-assisted content with real expertise, personality, and value ranks just fine. The distinction is quality, not authorship.
How many blog posts should I write?
Start with one per day and see what your capacity allows. Quality matters more than quantity — one genuinely useful post outperforms ten pieces of filler.
Can Claude Code submit my sitemap to Google Search Console?
Not directly — Google Search Console requires browser-based authentication. But Claude Code can generate the sitemap, set up automatic generation on deploy, and structure your site so Google finds everything.
Start here
If you want to try this workflow, start with one blog post. Pick a keyword with difficulty under 20, build a cluster, write the post with Claude Code, apply the on-page checklist. See how it feels.
Once you've done it manually and know what good looks like, then think about scaling. We're building SEO-specific skills that encode this entire workflow — keyword clustering, competitor analysis, content creation with voice injection, and on-page optimisation — into a single repeatable skill.