You have an SEO to-do list that never gets shorter.
Keyword research. Content creation. Technical audits. Sitemaps. Schema markup. Meta tags. Internal linking. Core Web Vitals.
Every single one of those tasks used to require either deep expertise, a full team, or both.
Claude Code changes that.
Built by Anthropic, Claude Code is an AI coding assistant that lives inside your terminal. It does not just write code. It builds entire websites, automates technical website auditing, generates SEO-optimized content, and handles the kind of production-level development work that used to take a team of specialists.
This guide will show you exactly how to use Claude Code for SEO, from setting up your first site to ranking for competitive keywords, without needing to know a single line of code.
Hi, I’m Brandon Leuangpaseuth. I run a fully remote growth SEO agency for B2B SaaS companies and venture-backed startups. Past clients include Keeper Tax (10K to 50K monthly visitors and a 700% conversion increase in 3.5 months), EasyLlama (991% organic growth in 9 months), and Joon App (500K+ monthly impressions in 2 months).
Everything in this guide is how I use Claude Code to build and execute SEO systems for clients. You can learn more about the Brandon Leuangpaseuth SEO agency and what we do.
What Claude Code Actually Is
A lot of people hear “coding assistant” and immediately check out.
Do not.
Claude Code is not just for developers. It is one of the most dependable AI tools available for anyone who wants to build and rank websites. You are the boss. Claude Code is the employee. You tell it what to do in plain English and it handles the execution.
Claude Opus is the most capable model in the Claude lineup, built by Anthropic as its flagship for tackling complex problems and enterprise-level development. Claude Cowork is the collaborative desktop version, useful for pairing with Claude Code on multi-step tasks like content production and backlink outreach. The web version rounds it all out for tasks you want to handle in a browser.
None of that matters much right now. What matters is understanding what Claude Code can do for your SEO results.
Here is the short version. Claude Code can:
- Build static, SEO-optimized websites from scratch using frameworks like Next.js
- Conduct automated technical audits and fix every issue it finds
- Generate schema markup, sitemaps, and robots.txt programmatically
- Write and optimize content following your brand voice and keyword targets
- Perform keyword research using live data from SEMrush via MCP integration
- Execute multi-step tasks end to end without constant supervision
That is not theoretical. SEO practitioners are already using it to do in 20 minutes what used to take full-time teams doing it all week.
Why Claude Code Matters for SEO Right Now
Google ranks websites on over 200 signals.
Nobody knows all of them. Not SEO agencies. Not SEMrush. Not even Google employees themselves.
But here is what we do know: the businesses ranking at the top of competitive searches are not doing it manually anymore. They have systems. Automated workflows. Optimized technical foundations.
Claude Code lets you build those same systems as one person.
Some companies like Homestars have entire divisions dedicated to on-page SEO and technical SEO full-time. With Claude Code, one person can replicate what those teams do, often faster, for a fraction of the cost.
That is the real opportunity here.
AI SEO is also exploding right now, and the good news is this: if you rank well in traditional search, you will rank well in AI search too. ChatGPT, Perplexity, and Google AI Mode all pull from the same high-authority sources that rank well on Google. One strong SEO foundation covers both.
Step 1: Build Your Website the Right Way from Day One
Before you write a single word of content, your technical foundation needs to be correct.
There are three types of website builds you will encounter when using Claude Code: static site generation, server-side rendering, and client-side rendering. Only one of them is right for SEO.
Think of it like ordering a pizza. Static site generation means the pizza is already made. You walk in, grab your slice, and you are gone in ten seconds. That is exactly what Google wants when it crawls your site: instant access to a complete, readable page so it can index and rank it.
Server-side rendering means you wait 20 minutes while they bake it. Client-side rendering means you have to make the pizza yourself and you never get it right. Google cannot rank what it cannot read.
You need static site generation. Every time.
To get started with Claude Code, create a new folder for your project. Inside that folder, create a file called CLAUDE.md. This is your operating manual. It tells Claude how your site should be built, what tone to use, what technical requirements to follow, and how to behave throughout the project.
Once that file is set up, Claude Code will follow those rules across every future task in that folder. Think of it like onboarding a new employee. Do it right once and you will never have to explain the basics again.
To build your site, open Claude Code and use a prompt like this:
“Please build a beautiful Next.js website for me with a homepage, a blog index page, and a services index page. Please build this with static site generation. Here is a screenshot of the design I want you to clone.”
Attach a screenshot from a site like Dribbble to give Claude a visual reference. Without one, the design will look generic. With one, it comes out polished on the first shot.
Your site will be live on a local server within minutes.
Step 2: Keyword Research with Real Data
Here is a mistake that kills most AI-assisted SEO projects before they start.
People ask Claude Code to generate a list of keywords directly from the AI without using any live data. Claude Code will give you its best guess. The problem is that its best guess does not include search volume, keyword difficulty, or competition level. You are flying blind.
The fix is integrating the SEMrush MCP into Claude Code.
An MCP (Model Context Protocol) is essentially a plugin that gives Claude Code access to live, real-world data. The SEMrush MCP connects Claude directly to SEMrush’s keyword database so it can pull accurate search volumes and competition scores before recommending anything.
To install it, run the SEMrush MCP command in your terminal and authenticate with your SEMrush API credentials. Once connected, Claude Code can run full research and data insights reports, not just guesses.
When filtering for winning keywords, use these three rules:
Keyword difficulty of 30 or below. Anything higher and a new site has almost no chance of ranking quickly. Search volume of at least 100 per month. Ranking for something two people search for is not worth the effort. Informational or commercial intent. Someone searching “how much does plumbing cost” is close to buying. Someone searching “what is plumbing” is probably a kid doing homework.
From there, build a topical authority map. This is the approach most serious SEOs use in 2026. Instead of publishing one article and hoping it ranks, you build clusters of related content around a central pillar page. A page about “Claude Code for SEO” becomes the center, with supporting articles covering keyword research, technical SEO, content creation, local SEO, and so on.
When Google sees 10 high-quality articles all pointing toward the same topic, it starts treating your site as an authority in that space. That authority lifts every page you publish.
Ask Claude Code to produce a 90-day topical authority roadmap using your SEMrush data. It will return a full content calendar with pillar pages, cluster articles, keyword targets, and an internal linking plan. Save that roadmap in your project folder. You will refer back to it throughout your build.
Step 3: Create Content That Actually Gets Read
Most AI content is unreadable.
Not because it is factually wrong. Because it reads like a brochure. Long, flat paragraphs. Generic sentence structures. No personality. Nobody wants to sit down on a Saturday afternoon and enjoy a plumbing blog post. That is just the reality.
But here is the thing about Google: it watches how people behave on your page. If someone lands on your article, reads three sentences, and bounces, Google takes note. If someone reads 90% of the article, scrolls back up, and clicks an internal link, Google takes note of that too.
High engagement signals push your rankings up. Low engagement signals pull them down.
So the goal is not just to write content. It is to write content people actually want to read.
The way to do this with Claude Code is context loading. Before Claude writes a single word, feed it reference material that teaches it how you write. This includes your voice, your opinions, your humor, your vocabulary, and any stories or case studies from your business.
Create a folder called “reference” inside your project. Inside it, create files like voice.md, humor.md, tone.md, and beliefs.md. Paste in LinkedIn posts, email transcripts, client call recordings, or any writing that sounds like you at your best. Then tell Claude to pull from those files whenever it creates content.
The difference between a cold AI draft and a context-loaded draft is dramatic. One sounds like a robot trying to be helpful. The other sounds like you actually showed up and wrote it.
On top of your voice files, Claude searches competitor content before writing each article. Give it the keyword, let it analyze the top three ranking pages, and instruct it to extract the average word count, number of H2 headings, image count, and topics covered. Then tell it to beat that average across every metric while writing in your voice.
That is the formula the top-ranking pages are using. There is no reason not to steal it.
Step 4: Automate Technical SEO
Technical SEO used to require coding experience.
Not anymore.
Claude Code enhances SEO performance through automated technical website auditing. You run a Google Lighthouse report, paste the results into Claude, and it fixes everything. No need to understand what “render-blocking requests” or “LCP request discovery” means. You paste, it fixes, you re-run the report.
The target is 100 out of 100 across performance, accessibility, best practices, and SEO.
Here is the exact process. Open Chrome DevTools (three dots, More Tools, Developer Tools). Go to the Lighthouse tab. Run an analysis on mobile. Copy the full report. Paste it into Claude Code and say:
“Here is my Lighthouse report. Please optimize my site to score 100 across all four categories.”
Claude Code will work through every issue. When it is done, re-run Lighthouse to check. If anything is still below 100, paste the new report back in and repeat. You should hit your targets within two to three iterations.
Beyond Lighthouse, Claude Code handles the three technical SEO essentials that are non-negotiable for ranking:
Sitemaps tell Google every page that exists on your site. Without one, pages can go unindexed for weeks. Claude Code can dynamically rebuild your sitemap.xml automatically every time you publish new content.
Robots.txt tells Google which pages it is allowed to crawl. You want it to crawl everything except your admin dashboard and any pages that would create duplicate content issues. Claude Code generates a clean robots.txt that handles this correctly.
Schema markup helps Google understand what your content is about at a structural level. Claude Code is proficient at generating clean, valid JSON-LD Schema for services, products, FAQs, and articles. Claude Code can programmatically generate schema markup for any page type in seconds, which makes your content eligible for rich snippets in search results.
Programmatically constructing FAQ structures, for example, can boost rich snippet eligibility significantly. Rich snippets take up more visual space in search results, which increases click-through rates even without changing your ranking position.
Structured data does more than help Google. It also improves how AI models retrieve and surface your content when answering questions. This is the core idea behind retrieval augmented generation, where AI pulls from indexed, well-structured sources to build its answers. If you want to understand how how retrieval augmented generation connects to your SEO strategy, that guide goes deep on it.
For Answer Engine Optimization, these structured data signals are especially important. AEO is about capturing featured snippets and AI-generated citations. The better structured your content, the more likely it is to get pulled into AI Overviews, Perplexity answers, and ChatGPT responses.
Step 5: Build Service Pages at Scale
Blog posts build topical authority.
Service pages make money.
The distinction matters. When someone searches “plumber Toronto,” they are not looking for a blog post. They are ready to hire. Service pages target those commercial-intent keywords, and they are the pages that directly convert visitors into paying customers.
Think of it like a zipper. On one side you have your services: emergency plumbing, drain cleaning, water heater repair. On the other side you have your locations: Toronto, Mississauga, Brampton, Scarborough. Zip them together and you get a matrix of service pages, one for each service-city combination.
Each of those pages is a separate shot at ranking. One page gives you one opportunity. Fifty pages give you fifty. Some will land immediately. Others will climb slowly over time as your domain authority grows.
To find the winning keywords for your service pages, go to SEMrush’s Keyword Magic Tool and sort by cost per click. High CPC keywords are keywords advertisers are actively paying for. That means there is proven commercial intent and proven revenue behind them. Those are the pages worth building.
Export your target keywords as a CSV file, drag it into Claude Code, and use a prompt like:
“Please build a service page for this keyword. Copy the layout from this screenshot. Make sure it uses static site generation.”
One prompt. One page. Repeat for every keyword in your list.
Keep the number of service pages reasonable. Google does not penalize you for having service pages. It does penalize you for having too many pages with nearly identical thin content. Somewhere between 10 and 100 is a smart starting range depending on your market.
Step 6: Package Everything Into a Reusable Skill
Here is where the real leverage kicks in.
You have built your site, done your keyword research, generated your first blog post, optimized it for on-page SEO, and run the technical audit. That whole process probably took you a few hours the first time.
Now imagine doing it every single day.
Claude Code lets you package your entire workflow into a single command called a skill. A skill is a reusable template that runs your full production process from one word.
Type “/blog” and Claude Code will: pull the next unwritten keyword from your content calendar, research the top three competing articles for that keyword, write a full article in your voice, add keyword clusters for secondary and tertiary keywords, optimize the article for on-page SEO, update your sitemap, and log the post as complete so it never writes about the same keyword twice.
Type “/service” and it builds an entire optimized landing page.
Type “/review” and it generates a month of Google Business Profile posts.
Set that skill to run on a scheduled task every morning at 9 AM and your content operation runs itself.
This is not a future promise. Scheduled tasks are available in Claude Code today. You can set them up under the desktop app’s task manager and point them at a GitHub repository. Every time Claude Code writes a new piece of content, it pushes it to GitHub. Netlify or Vercel detects the new commit and deploys it automatically.
One person with Claude Code running 24/7 will outpace a team of 20 manual SEO writers. Not slightly. By a significant margin.
The one thing you must not do is flood your site with content all at once. Google watches publication velocity. A spike from zero posts to 200 posts overnight looks like spam. Start slow. One post a day. Increase gradually over weeks. Slow and steady here is not caution, it is strategy.
Step 7: Deploy Your Site and Get It Indexed
A site sitting on your local machine is invisible to Google.
Deploying takes about 15 minutes and it is free.
The process is three steps. First, push your code to GitHub. GitHub is like Google Drive for code. Create a free account, make a new private repository, copy the setup code GitHub gives you, and paste it into Claude Code with the instruction:
“Please push my entire project to GitHub.”
Second, connect GitHub to Vercel or Netlify. Both are free for the tier you need. Import your project, confirm the framework is set to Next.js, and click deploy. Your site will be live in under 60 seconds with a public URL.
Third, submit your site to Google Search Console. This is how Google officially learns your site exists. Copy your site URL into Search Console, verify ownership using the HTML meta tag method (just paste the tag into Claude Code and ask it to add it and redeploy), and then submit your sitemap.xml URL in the Sitemap section.
Once your sitemap is submitted, every page you publish will be tracked and indexed by Google. You can also use Search Console’s URL Inspection tool to manually request indexing for new pages. Google typically processes these within 24 hours instead of the weeks it can take otherwise. You get roughly 10 manual indexing requests per day, so use them strategically on your most important new pages.
Set up Google Analytics at the same time. It will track how visitors behave on your site, which pages drive engagement, and which pages are bleeding traffic.
Building for AI Search, Not Just Google
Here is something most SEO guides are not talking about yet.
Google is not the only place your content needs to rank anymore.
AI search tools like Perplexity, ChatGPT, and Google AI Mode are all pulling content from the web to answer user questions. The brands that show up in those AI-generated answers are getting cited, mentioned, and recommended to people who never even open a traditional search result.
The way you win in AI search is through LLM SEO, or getting your brand cited and recommended by large language models.
The process has four steps.
First, identify the prompts you want your brand to appear for. Not just “best SEO software” but every variation of that intent: “what AI tool helps agencies with SEO,” “I need to automate internal linking for my clients,” “best tools for scaling content production.” The intent matters more than the exact phrasing.
One tactic worth understanding before you start producing content for AI search is seeding. This is the process of deliberately getting your brand and content into the training data and citation pools that LLMs pull from. This LLM seeding strategy guide breaks down exactly how it works.
Second, check what is currently being cited when those prompts are asked. Look at the cited sources. Are they listicles? Comparison pages? How-to articles? That content format is what the AI has decided it trusts for that topic. You need to produce the same type of content, but better.
Understanding what makes AI models trust and cite certain content over others is its own discipline. The signals are different from traditional Google ranking factors in some important ways. This breakdown of LLM ranking factors is worth reading before you optimize for AI search.
Third, produce a more complete, better-structured version of the content that is currently being cited. Claude Code makes this fast. Feed it the competitor URL, your keyword targets, and your brand voice files, and ask it to produce a superior version.
Fourth, track whether your content is being cited. Tools like PromptWatch let you monitor which AI platforms are mentioning your brand and with what sentiment. There are also specific tactics for improving your chances of getting cited by LLMs that go beyond just publishing good content, and that guide covers them step by step.
The brands winning in AI search are not doing anything mystical. They are producing the best available answer to the questions their customers are asking. Claude Code makes producing that content faster than anything else available right now.
On-Page SEO: The Checklist Claude Runs for You
On-page SEO is the part most people either skip or get wrong.
There are more than 80 individual signals Google looks at on a page level. Things like whether your primary keyword appears in the first 100 words. Whether you have exactly one H1. Whether you have the right ratio of H2s to word count. Whether your meta title is between 50 and 60 characters. Whether your meta description is between 150 and 160 characters. Whether you have the right number of internal and external links.
Claude Code handles all 80-plus signals in a single pass.
After your draft is written, paste your full on-page SEO checklist into Claude Code and say:
“Please update this blog post to meet every item on this checklist while maintaining the voice and style of the original.”
That last part is important. You do not want Claude to turn your personality-driven article into a dry, keyword-stuffed document. The goal is to optimize for structure and signals while keeping the writing engaging.
Unique title tags should be between 50 and 60 characters. Unique meta descriptions should be between 150 and 160 characters. Never use duplicate title tags or meta descriptions across your site. These basics alone will put you ahead of most sites Google sees.
Getting Backlinks Without Doing It the Boring Way
Backlinks are still a major ranking signal.
The bad news is there is no fully automated way to build quality links yet. The good news is Claude Cowork makes the process significantly less painful.
Claude Cowork is a desktop collaboration tool that works alongside Claude Code on longer, multi-step tasks. For backlinks, the workflow is simple: you give Claude Cowork a list of target sites, directories, and guest posting opportunities. It handles the writing, form-filling, description crafting, and button-clicking. You just show up to log in where required.
That workflow can produce seven or eight quality links in 30 minutes, compared to the same amount of tedious clicking that used to eat an entire afternoon.
There are four backlink strategies worth pursuing:
Broken link outreach involves finding links on competitor sites that no longer work and offering your content as a replacement. SEMrush’s site audit tool surfaces these quickly.
One of the most underrated ways to earn links naturally is building content people want to reference. Guides, original data, and free tools all attract links without any outreach at all. If you want a deeper breakdown of how to build those, this guide on linkable assets for SEO covers the full strategy.
Guest posting means finding sites in your niche that accept contributor content, writing an article for them, and including a link back to your site. Search Google for your niche plus “write for us” to find opportunities.
Digital PR means getting mentioned in news articles, roundups, and industry publications. Being cited by a high-authority publication passes more link equity than a dozen low-quality directory links.
Startup and AI directories are low-hanging fruit for new sites. These directories accept listings for free and provide immediate backlinks that help with initial indexing.
What to avoid: paid link farms, private blog networks, and any service offering 100 backlinks for $5. These work briefly, get detected by Google, and can permanently damage your domain.
Off to the Races
Claude Code for SEO is not a shortcut.
It is a system.
The difference between sites that rank and sites that disappear is not who publishes the most content. It is who builds the strongest foundation, produces the most useful articles, and maintains consistent quality over time.
Claude Code removes the bottlenecks that used to make that impossible for a small team or a solo operator. The keyword research, the technical audits, the content production, the on-page optimization, the deployment, and the indexing can all run with minimal manual intervention.
The strategy still matters. The content quality still matters. The voice still matters.
But now you have an AI coding assistant that handles every technical piece, so you can focus on the strategic decisions that actually move the needle.
If you run a B2B SaaS company, a startup, enterprise or a growing business and you want an SEO system built from the ground up to rank and convert, apply to work together here.
