Marketing Systems

Claude Marketing Skills: Corey Haines Builds a Growth Engine

Direct Answer

Corey Haines's marketing skills turn Claude Code from a generic copy assistant into an operator with repeatable processes for site building, SEO, email, analytics, and acquisition. In Andrew Warner's demonstration, the starting point is a single-page AI concierge business called Setup Claw. Corey uses an agent and his skill library to turn that starting point into a more complete marketing system.

The striking claim that years of marketing work can fit into a week is Corey's assessment of execution speed, not an observed revenue result from this business. A generated landing page, page plan, sequence, or audit is an artifact. Customers, rankings, qualified leads, and profitable campaigns still have to be earned and measured.

The useful shift: keep the offer, customer evidence, brand rules, and measurement definitions in shared context. Let agents perform bounded implementation work, then review the live customer experience and business signal.

Watch the Episode

Credit: This article interprets Andrew Warner's interview with Corey Haines, published 11 February 2026. Corey founded Conversion Factory. The demo shows work in progress, not independently verified acquisition results.

What the Skills Actually Do

Corey's marketing skills repository is a library of reusable agent instructions. Instead of asking a general model to “do marketing,” a founder can invoke a focused process for conversion copy, SEO, email, measurement, or another job. The repository is the current installation reference; its contents and supported tools may have changed since the recording.

The skill is not a substitute for a brief. It shapes the method, while the founder supplies the audience, promise, proof, constraints, and permission to change the site. The video shows Claude Code creating files and working inside a Next.js project. That ability is useful precisely because it can change real infrastructure, and it deserves a review gate before deployment.

Begin With Product-Marketing Context

Midway through the demo, Corey points out that the agent should know the product before generating downstream assets. A product-marketing context file can capture who the service is for, which problems it solves, what it does not promise, what proof exists, the desired action, and the voice to use. Then each skill can read the same facts rather than improvising a slightly different company every time.

For Setup Claw, the initial site presents an AI concierge service. The editorial question is not “how many pages can Claude make?” It is which buyer, use case, and offer deserve a page, and whether someone would trust it enough to contact the business.

From a One-Page Site to a Reviewable Funnel

Corey starts by using Claude Code to build a fuller website and applies the marketing skills to rewrite the homepage. He works on the value proposition, landing-page structure, and calls to action. The advantage is a shared workflow: the same agent can edit the copy and the underlying page instead of leaving a marketer to hand off a document to an engineer.

Before shipping, review the page on mobile, test every form and destination, check accessibility and load behavior, and have the owner approve any testimonial, guarantee, comparison, or numerical claim. A polished preview is not a validated funnel.

Programmatic SEO: Scale the Question, Not Just the Template

The episode's programmatic SEO discussion starts with pages for specific integrations and expands into combinations of tools and use cases. Templates can make this affordable: a founder defines a useful page pattern, then the agent fills it from structured, verified inputs. The number of possible combinations grows quickly, but that arithmetic says nothing about demand or quality.

Start with a short list of real questions buyers ask. Each page needs a distinct answer, working examples, accurate integration details, and a clear route to the product. Index only pages that genuinely help someone complete a task; use analytics and search data to decide what to improve or remove. Publishing hundreds of mechanically varied pages is not a growth strategy.

Nurture Email: Build a Sequence Around One Decision

Corey asks the agent to create an email nurture sequence and trims it from eight emails to six. That small exchange matters: the agent can draft quickly, but a human still decides the number of messages, what each one contributes, and when to stop. A useful sequence moves from a subscriber's problem to proof, objections, next steps, and a clear offer without repeating the same pitch six times.

Consent and list source matter as much as wording. Review links, unsubscribe behavior, sender reputation, segmentation, and any automated follow-up rule. Then measure replies and qualified conversions, not just generated email volume.

Audit the Site, Then Connect Analytics and Ads

The demo's SEO audit surfaces technical checks such as crawl access, sitemap, metadata, canonical URLs, structured data, and social previews. Some fixes are straightforward for an agent working in the repository. Others require product judgment or evidence that the content deserves to rank. Corey's comparison with expensive agency audits is a provocative framing, not proof that automated audits replace strategic SEO work.

Analytics and attribution close the loop: record which page, campaign, or search query brings qualified interest. Corey also discusses paid ads that an agent can monitor and adjust. For real accounts, begin with read-only analysis and human-approved recommendations. Set a budget ceiling, change log, approval threshold, and rollback plan before granting write access to ad platforms.

A Responsible First Run

StepAgent outputHuman decisionProof to keep
1. ContextDraft offer and customer briefApprove claims, evidence, and boundariesVersioned brief
2. PageOne focused landing-page changeTest the buyer flow on desktop and mobilePreview, diff, form result
3. SEOOne distinct use-case page and technical auditCheck accuracy and usefulness before indexingSources and crawl checks
4. NurtureShort sequence for opted-in leadsApprove timing, links, and consentRendered emails and test sends
5. LearnWeekly qualified-lead reportChoose one experiment; authorize any spendBaseline, result, and correction

The video uses a permissive agent setting to move quickly through a demonstration. Do not copy that permission level into a live business environment. Use a branch or staging site, narrow tool access, and approval before publishing, sending email, or spending money.

Video Chapters

TimeTopicTimeTopic
00:0010x more marketing with AI12:09Growth via tool combinations
00:18Corey Haines and Conversion Factory13:57Email nurture sequences
00:36Building versus getting users18:18Why six emails, not eight
00:45Turning a side project into revenue19:21Product-marketing context
01:21Limits of a single-page website21:18Automated SEO audit
02:06Claude Code setup22:57Agency SEO costs
04:12Building a landing page24:09Marketing skills overview
05:33Installing the marketing skills25:48Connecting analytics
06:54Structured homepage copy27:00Managing paid ads
08:15Programmatic SEO29:06Technical SEO fixes
10:12Integration pages at scale30:00Marketing plus vibe coding

Verdict

The compelling part is not that an agent can make many pages in minutes. It is that one founder can preserve a consistent product brief and use focused skills to connect site changes, content, email, SEO, and measurement. This can collapse handoffs and make small experiments cheaper.

It only becomes a growth engine when the outputs survive review and the business measures who showed up, what they needed, and whether they converted. Use the speed to run better tests, not to flood channels with unverified assets.

Sources and Links

Publication date follows the primary video's YouTube date, 11 February 2026. Editorial review: 16 September 2026. The repository and third-party tool features may have changed since the recording.

Common questions

What are Corey Haines's marketing skills?
They are reusable instructions for AI coding agents covering marketing tasks such as positioning, copy, conversion, SEO, email, analytics, and paid acquisition. The current GitHub repository is the source for installation and supported agents.
Do I need to be a developer to use the skills?
You need to be able to define an offer, provide accurate customer context, review changes and claims, and test the result. Technical work is easier with an agent, but the site, integrations, tracking, and publishing still require careful review.
Will programmatic SEO pages automatically rank?
No. A generated page is only a draft. Each page needs a distinct user need, accurate data, original value, internal links, and performance monitoring. Mass-producing near-duplicates can create thin content rather than useful search coverage.
Can an AI SEO audit replace an agency?
It can surface and sometimes fix routine technical problems. It cannot by itself validate market positioning, information quality, search demand, implementation impact, or revenue attribution. Corey's agency-cost comparison is his opinion, not a measured replacement study.
What should I measure first?
Start with one defined audience and offer. Track qualified visits, meaningful form submissions, booked calls, conversion rate, and source quality; review consent, deliverability, and ad-spend limits before automation.
Share
X LinkedIn Reddit
Build Yours

Want a system
like this one?

Book a free 30-minute call. We map your situation, identify the highest-impact automation, and figure out if we are a fit.

Book Free 30-min Call