How I Audited My Blog's SEO with Claude Code

Auditing a blog with Claude Code is a faster way to surface problems with structure, metadata, internal links, slugs, duplicate content, and AEO opportunities. It won't replace a professional SEO audit, but it does help you document patterns and turn findings into actionable tasks.

Why did I audit my own blog with Claude Code?

I did it because I wanted to review the blog as a system, not as a loose collection of posts. Reading a single article and deciding "this is fine" is one thing. Checking whether every article shares consistent metadata, clean URLs, internal links, AEO structure, and refresh opportunities is something else entirely.

Read this as a documented process case study, not a promise of magic results. AI can speed up an audit, but it doesn't replace interpretation. Claude Code can walk through files, find patterns, compare structures, and suggest changes. But deciding what matters, what ships, and what gets prioritized is still strategic work.

My goal was to answer concrete questions: which articles are ready to compete? Which ones fell short? Which slugs are unclear? Where are the FAQs missing? Which pieces should link to each other? Which topics could grow into clusters? You don't solve that kind of question with a "general analysis." You solve it with systematic review.

What I found interesting is that Claude Code helped me see the blog as an editorial product. Not just as content. And once you see a blog as a product, you start noticing technical debt, information debt, and ranking debt.

What would I review first in an AI-assisted SEO audit?

First, I'd review anything that can keep the content from being understood: titles, meta descriptions, H1s, slugs, categories, internal links, heading structure, and a direct answer to search intent. After that, I'd review depth, authority, and differentiation.

Google explains that SEO helps search engines understand content and helps users find a site through search ¹. That definition matters even more once you also want answer engines to interpret the content. It's not enough for an article to exist. It needs clear signals about topic, intent, authorship, and usefulness.

My first layer of auditing looks like this:

Claude Code helps because it can review many files against the same criteria. The human helps because they know whether those criteria actually serve the brand's positioning.

How did I prepare the audit before asking Claude Code for anything?

I prepared the audit by defining a checklist, a taxonomy of problems, and the expected output format. If you don't tell the AI what to look for and how to report it, it'll hand back useful observations buried in noise.

This part is key. A lot of people say "audit my blog" and expect a brilliant answer. But a good audit needs scope. Reviewing technical SEO, content, AEO, conversion, internal links, or performance are not the same job. All of them can matter, but you don't review them the same way.

Before running any analysis, I defined three tiers:

1. Critical problems. Anything affecting comprehension, indexing, URLs, duplicates, or base structure.
2. Editorial improvements. Weak answers, vague titles, missing examples, thin authority, or incomplete sections.
3. Strategic opportunities. New clusters, bridge articles, comparisons, FAQs, and internal links.

Then I asked for a tabular output. That keeps the analysis from turning into an inspirational paragraph. An audit has to become a backlog. If you can't assign it, prioritize it, or execute it, it isn't an audit yet; it's an opinion.

What prompts did I use to audit the blog?

I used prompts that separate exploration, diagnosis, and actions. First I asked for an inventory, then an evaluation, then prioritization. That sequence keeps the AI from proposing changes before it understands the system.

A useful first prompt would be:

Review the structure of this blog as an editorial system. Build an inventory of articles, slugs, categories, SEO titles, meta descriptions, H1s, number of H2s, presence of an FAQ, internal links, and CTA. Don't propose any changes yet.

Next comes the diagnosis:

Based on the inventory, identify SEO and AEO problems. Classify each finding as critical, medium, or low. Explain why it matters and what concrete action you recommend.

And finally the prioritization:

Turn the findings into an improvement backlog. Order by SEO/AEO impact, implementation difficulty, and relevance to UX/UI design, Webflow, AI, neuromarketing, branding, and paid media services.

Claude Code works especially well for this kind of task because it can operate inside the project, read files, search for patterns, and synthesize findings. Anthropic describes Claude Code as able to automate development tasks, work with files, run commands, and understand entire codebases ². For a static or file-based blog, that capability becomes genuinely useful.

What findings would I look for in a blog like blog.israelpina.cool?

I'd check whether the blog builds authority around clear topics: AI-assisted design, Webflow, UX/UI, neuromarketing, branding, and paid media. I'd also check whether each article connects to services, case studies, or resources that can turn a read into a conversation.

This isn't about publishing for the sake of publishing. It's about building clusters. If a blog has an article on AEO, it should connect to llms.txt, JSON-LD, citability in ChatGPT, SEO audits, and Webflow. If it has articles on AI-assisted design, it should connect Figma, Claude, prompts, prototyping, ethics, and workflow.

The audit has to answer whether those pillars are alive or just show up as categories. A category with no internal links, no pillar articles, and no clear CTA is a label, not a strategy.

How do I turn findings into concrete actions?

I turn findings into actions using an impact-versus-effort matrix. Not everything you find has to be fixed right away. The improvements that boost clarity, internal linking, and conversion without requiring huge rewrites come first.

My simple matrix looks like this:

AI-assisted auditing shines when it turns repetitive problems into repeatable tasks. For example: "every AEO article must open with a direct definition," "every Webflow article must link to services," "every post must have an FAQ and a soft CTA," "every cluster must have a pillar article."

That kind of editorial rule improves the blog as a system. And when the system improves, every new article inherits better conditions.

What mistakes would I avoid when auditing SEO with Claude Code?

I'd avoid accepting recommendations without verifying them, rewriting everything out of anxiety, optimizing for keywords alone, and confusing the number of findings with the quality of the audit. A good audit prioritizes; a bad one overwhelms.

The most common mistake is letting the AI propose endless changes. There will always be something to improve. But not everything improves the business. If the goal is to attract clients for design, Webflow, or digital strategy, prioritization has to connect to those services. An article can be "SEO-correct" and still not move the needle.

Another mistake is applying automation without editorial control. If you update titles, meta descriptions, and internal links in bulk, review tone consistency. The AI can write a functional meta description, but it may not sound like your brand. And on a personal-professional blog, voice matters.

I'd also avoid measuring success by traffic alone. For a consultant or studio, quality is what counts: the right inquiries, perceived authority, time on page, internal links followed, qualified leads, and the ability to be cited.

Checklist to audit your blog with Claude Code

Before closing out the audit, I'd review the following:

1. There's a complete inventory of articles and metadata.
2. Every article has a clear, unique H1.
3. Every meta description promises a concrete answer.
4. The slugs are readable and on-topic.
5. The H2s answer real questions.
6. There are internal links between articles in the same cluster.
7. The FAQs answer long-tail questions, not filler.
8. Every article has a soft CTA tied to services.
9. The pillar articles are identified.
10. The findings are prioritized by impact and effort.

The audit ends when you have an actionable plan. If all you've got is a long document, you still have to turn it into decisions.

Frequently asked questions

Can Claude Code run a complete SEO audit?

It can help a lot with inventory, patterns, structure, and recommendations, but it doesn't replace strategic analysis, keyword research, Search Console data, or editorial judgment.

Does my blog need to be code-based to use Claude Code?

It works best when the blog lives in files or a repository. If it's on a closed CMS, you can still use it with exports, sitemaps, scraping, or documentation, but the workflow changes.

What's the difference between auditing SEO and AEO?

SEO reviews visibility in search engines; AEO reviews whether the content answers clearly and can be used by answer engines. They complement each other.

Can I automate changes across all my posts?

You can, but it's worth doing with review. Automating without control can break tone, links, formatting, or metadata.

What would I review first if I'm short on time?

I'd review titles, metas, H1s, the direct answer up top, internal links, and CTAs. Those are high-impact changes with relatively low effort.

Soft CTA

If you want to turn your blog into a clearer editorial system for SEO, AEO, and conversion, email me at hola@israelpinapol.com or visit israelpina.cool. AI can speed up the audit, but strategy decides what's worth fixing.

---

1. Google Search Central, "Search Engine Optimization Starter Guide," consulted for an operational definition of SEO, helpful content, and discovery: https://developers.google.com/search/docs/fundamentals/seo-starter-guide ↩
2. Anthropic Docs, "Claude Code Overview," consulted for capabilities around reading a codebase, editing files, running commands, and automating tasks: https://docs.anthropic.com/en/docs/claude-code/overview ↩