--- title: "Claude Code for Marketing: Content, SEO and Site Work | ONEGTMLAB" description: "How we use Claude Code for marketing: a LinkedIn content engine, SEO audits, a site migration, link prospecting, page research and lead magnets." canonical: https://www.onegtmlab.dev/growth-playbooks/claude-code-for-marketing --- # Claude Code for Marketing: Content, SEO and Site Work | ONEGTMLAB [![](/_astro/69a56255e2154ea94c75641b_onegtm-nav-logo.CkLAn8xX.svg)](/) [ Book a call ](/book) Growth Playbook # Claude Code for Marketing The same tool that runs our outbound also runs our marketing: the LinkedIn content engine, the SEO work on this site, research before a page is written, link prospecting and gated lead magnets. ![](/_astro/69a8569f8db69df76ba584e1_jainendra-ojha.CeNJqXRl_Z11dhei.webp) By [Jainendra Ojha](https://www.linkedin.com/in/jaizendra/), Founder, ONEGTMLABLast updated 25 Oct 2026 [ Book a Strategy Call ](/book)[ All playbooks ](/growth-playbooks) What it is ## What it is and when to use it Claude Code reads and writes files, runs scripts and calls APIs, which makes it useful for marketing work that is part writing and part data: measuring what a good post looks like, inventorying every URL on a site, checking every claim on a page against its source, or comparing hundreds of competitor pages without reading them by hand. We keep each job as a skill with its own checks, so the work is repeatable and the lessons stay with it. This website is an example: it was moved off its old builder and rebuilt with Claude Code, and every change is gated by a speed test and a check of every URL. ### Use this playbook when - A founder wants to post on LinkedIn consistently in their own voice, without the posts all sounding the same. - A website needs an SEO audit, a migration or new pages, and every change must be proven not to break anything. - You are about to write a service or comparison page and need to know what the pages that already rank cover, and what they leave out. - You want links from relevant sites, or a gated lead magnet that captures work emails. Tools ## Tools we use Only tools we have run this playbook with, and what each one does in it. - Claude Code: Skills for content, SEO, research and page builds; subagents for reading at volume - Apify: Scrapes of LinkedIn posts used as measured references - Firecrawl: Site maps and clean page scrapes - Serper: Google results for link prospecting searches - DataForSEO and Ahrefs: Search volume and keyword checks before a page or name is chosen - NotebookLM: An independent reader for a large scraped corpus - Astro and Cloudflare: This website: static pages, Markdown copies and hosting - Lighthouse: Speed and quality gate on every page change How it works ## The steps The order we work in. Each step feeds the next, and every result is written to the database as it happens. 1. 01 ### Give the content engine a brain Positioning, ideal customer, voice rules and banned phrases live in one folder the model reads first. Each content type gets its own skill, and every draft is saved with a date. 2. 02 ### Measure reference posts, never copy them We scrape a sample of recent posts from reference profiles and measure them: length, line count, hooks, closers and structure. Drafts are calibrated to those measurements, and the wording is always our own. 3. 03 ### Check every tool before naming it Before a post names a tool, its job category is checked against a reference map, so a sequencer is never described as a data source. 4. 04 ### Inventory the whole site For any site job we list every URL first: a full site map at the maximum limit, plus the sitemaps and llms.txt the site publishes, merged and de-duplicated. 5. 05 ### Audit, fix and gate SEO audit skills find the problems. Every fix ships only after a Lighthouse run and a check of every URL in the site map. 6. 06 ### Research a page before writing it The competitor set comes from real search results. Their sites are mapped, page types are counted by script, subagents read the pages verbatim and their quotes are checked. Then we write, and grep every phrase against the competitor text before shipping. 7. 07 ### Prospect links in bulk, judge them by meaning Footprint searches run deep into the results, junk hosts are filtered out, subagents keep or drop each site by what it is, and a probe confirms the link would be followed. 8. 08 ### Choose names and topics by measured demand Page names, service names and topics are checked against real search volume before they are chosen. 9. 09 ### Ship lead magnets with one job each A gated page with proof, a work-email form, a deck behind it and one booking call to action. Good output ## What good output looks like Measured results from this work: - This site's move to its new stack: median mobile Lighthouse performance across 13 pages went from 73 to 99, largest contentful paint from 4.09 s to 1.64 s and page weight from 3.20 MiB to 0.26 MiB. - Link prospecting: running the same searches seven result pages deep returned 12,992 domains for $4.82, against 2,735 domains at two pages. Judged by meaning, 13% were worth keeping. - Page research for a service page: 137 competitors and 81,438 URLs mapped and analyzed, for about $1.50 of external spend. - On our co-founder's last 47 LinkedIn posts, the 5 lead-magnet posts averaged 150.2 comments, against 41.95 for the other 42. Common mistakes ## Common mistakes, and the fix Each of these happened on a real run. The fix is now part of how we work. 1. A site map returned exactly 100 URLs. Cause The mapping call used the tool's default limit. Fix Always send the maximum limit, and re-map by section if the result hits the ceiling. 2. A search API key returned zero results without an error. Cause The free key silently ignored quoted phrases. Fix Test one quoted query on every key before a run, and never swallow API errors. 3. The junk-host filter removed 47 real domains. Cause The "random-looking domain" rule was too broad. Fix Tighten the rule and ship it with tests that include real domains. 4. A "free" directory listing turned out to be paid. Cause On many directories, the followed link is the product being sold. Fix Check for a paywall and probe the link before filling in any form. 5. A finding of "zero mentions" on competitor pages was wrong. Cause The search only covered one region of each page. Fix State the scope of every count, or read the full page. 6. Our draft copy echoed a competitor's sentence. Cause The phrase check ran once, before later edits. Fix Grep every phrase against the competitor corpus after every edit. 7. Counts from an AI notebook were wrong. Cause The notebook retrieves a subset of sources and extrapolates. Fix Take quotes from the notebook and counts from a script. 8. A post named a tool in the wrong role. Cause No check of what the tool actually does. Fix Check the tool context map before drafting. FAQs ## Questions people ask about claude code for marketing Is Claude Code good for marketing? For marketing work that mixes writing with data, yes. We use it for a LinkedIn content engine that reads positioning and voice files first, for SEO audits and a full site migration, for research on hundreds of competitor pages before writing one, for link prospecting and for lead magnets. It is weakest when nobody checks its output, so every job has its own checks. How do you use Claude Code for SEO? We start by listing every URL on the site, then run audit skills, fix what they find and gate every change with Lighthouse and a check of every URL. Before writing a page we map the pages that already rank and check our copy against them, and we pick names and topics by measured search volume. On this site that work took median mobile performance from 73 to 99. What is the difference between Claude and Claude Code? Claude is the chat assistant. Claude Code runs Claude as an agent in a terminal, where it can read and write files, run scripts and call APIs. That is what lets it inventory a website, scrape and measure reference posts or rebuild pages, rather than only draft text. ## Related service - [Content Engineering LinkedIn content built around your founder's voice that generates demand.](/linkedin-personal-branding) - [GTM Engineering Signal-led outbound systems that enrich, score, sequence and sync to your CRM.](/ai-gtm-engineering) ## Related playbooks - [LinkedIn Content Strategy Positioning first, measured structure, varied posts, and commenters captured as qualified signals.](/growth-playbooks/linkedin-content-strategy) - [Claude Code for Sales Run list builds, qualification, outreach and CRM sync from one terminal, with every judgment written down.](/growth-playbooks/claude-code-for-sales) - [Social Listening Playbook Turn LinkedIn posts, event chatter and competitor engagement into a qualified list, not a noisy feed.](/growth-playbooks/social-listening) ## Further reading - [How To Build A LinkedIn Content Engine Using Claude Code In 2026?→](/blog/how-to-build-a-linkedin-content-engine-using-claudes-code-in-2026) - [The Automated LinkedIn Content Distribution System For B2B Business→](/blog/the-automated-linkedin-content-distribution-system-for-b2b-business) - [The Signal-Based Outbound System→](/gtm-stack) - [GTM Glossary: the terms used in these playbooks→](/glossary) GET STARTED ## Want this playbook running on your market? Tell us what you sell and who you sell to. We will show you how this playbook would run for you, and what we would check first. [ Book a Strategy Call ](/book)[ See Case Studies ](/case-studies) [![](/_astro/69a56255e2154ea94c75641b_onegtm-nav-logo.CkLAn8xX.svg)](/) COMPANY [ Why Us ](/#why-us)[ Process ](/#timeline)[ Blogs ](/blogs)[ About ](/about)[ Clients ](/clients)[ Free resources ](/free-resources)[ Newsletter ](/newsletter) Services [ GTM Engineering ](/ai-gtm-engineering)[ Account-Based Marketing ](/services/account-based-marketing)[ Paid Media ](/services/paid-media)[ Content Engineering ](/linkedin-personal-branding)[ RevOps ](/services/revops)[ SEO, AEO & GEO ](/services/seo-aeo-geo)[ All services ](/services) © 2026 ONEGTMLAB Private Limited. All rights reserved. ![](/_astro/wordmark-3d.m-LgRZCQ.svg)