The Vibe Marketing Playbook

The philosophy

Everyone's selling hype. You're selling compound growth. The boring angle is incredibly differentiated.

If you've spent any time using AI for marketing, you've probably noticed something frustrating: everyone has the same tools, the same prompting guides, and output that sounds the same—generic headlines, corporate-speak copy, email sequences that read like they were written by a committee of robots who've never actually sold anything.

You've tried better prompts, longer prompts, prompt chains and templates. The output is still fine. Just fine.

Here's the insight: the problem isn't your prompts. Prompts are instructions. What you need is methodology.

The core insights

The three layers

Vibe Marketing stacks three layers that most people try to replace with prompts alone.

The complete tool stack

Claude Code setup

Install skills to ~/.claude/skills/—they load automatically. One-time setup, permanent improvement.

# Install Claude Code
npm install -g @anthropic-ai/claude-code

# Create skills directory
mkdir -p ~/.claude/skills

# Move your skill files there
mv ~/Downloads/skills/*.md ~/.claude/skills/

# Start Claude Code
claude

About skills

What is a skill?

A skill is an instruction manual that teaches AI how to do a specific task—what to look for, how to structure output, and which questions to ask—so you don't have to invent the workflow each time.

What's inside these skills

Research, frameworks, and examples of what good looks like, so you get professional-grade output without needing to be a marketing expert.

The ten marketing skills

Every skill shares brand memory, chains into workflows, and stacks on complex projects.

Strategy skills

The foundation skills that inform everything else.

Copy & content skills

Creative Engine

One skill, five production modes. It picks the right model (via Replicate) and shares a brand kit across modes. Note: requires a Replicate API token (pay-per-gen). The ten marketing skills work without API keys.

Initial research framework

The mistake most people make is not spending enough time in research. Cast a wide net, gather deep context, let AI work with real information.

The golden rule: cast a wide net, gather deep context, then let AI work.

Research prompt template

We are creating a [TYPE OF BUSINESS]
Example: "AI consulting agency for B2B SaaS"

Targeting: [AUDIENCE + REVENUE RANGE + INDUSTRY]
Example: "$2-10M SMBs in fintech who struggle with AI adoption"

What makes us different: [YOUR UNIQUE APPROACH/METHODOLOGY]
Example: "We use a confidently boring, systems-first approach
leveraging Claude Skills + MCPs vs. typical AI hype"

Current state: [Starting point]
Example: "No online presence. Starting from scratch."

Research Goals:
• Market landscape & key competitors
• Customer pain points & positioning gaps
• Pricing models & service packaging
• Enterprise best practices to repackage for our target

Tools for research

• Perplexity MCP — deep market research
• Web search — quick competitor lookups
• Firecrawl — scrape competitor sites at scale
• Playwright — screenshots and UX patterns

Expert review framework

You built something with AI—is it actually good, or just polished? Task-based agents add independent perspectives.

How to invoke expert review

"Spin up task-based agents to review this:
1. Agency Growth Expert
2. SEO Specialist
3. Conversion Expert
4. [Industry-specific expert]

Each analyzes from their perspective.
Synthesize: Where do they agree?"

When to checkpoint: after copy (before design), after design (before launch), after strategy (before execution), before any high-stakes decision.

The five-stage build sequence

Research → Foundation → Structure → Assets → Iteration

1. Research: deep context

   Market research, competitor analysis (including screenshots), and customer language before you write.

2. Foundation: voice + positioning

   Brand voice extraction, three to five positioning angles, explicit anti-positioning.

3. Structure: keywords + pillars

   Keyword gaps, content pillars, quick wins for the next sixty to ninety days.

4. Assets: pages, emails, content

   Landing copy, sequences, lead magnets—where methodology shows up in the work.

5. Iteration: reject until right

   Rejection cycles, voice enforcement, quality gates—would you actually ship this?

Decision framework

Six methods when research gives you too many options.

Skill stacks

Foundation stack

“Know who you are.”

Research (Perplexity MCP) → Market landscape, competitor gaps
        ↓
Brand Voice Skill → Voice profile, tone guide
        ↓
Positioning Angles Skill → 3-5 angles, differentiation

Output: voice profile, positioning angles, clear differentiation.

Conversion stack

“Turn visitors into customers.”

Direct Response Copy Skill → Landing page copy, headlines
        ↓
Frontend Design Skill → Production-ready design
        ↓
Lead Magnet Skill → Opt-in offer, bridge to paid

Traffic stack

“Get discovered.”

Keyword Research Skill → Keyword strategy, gaps
        ↓
SEO Content Skill → Ranking content
        ↓
Content Atomizer Skill → 15+ assets per piece

Nurture stack

“Build relationship over time.”

Email Sequences Skill → Welcome series, automation
        ↓
Newsletter Skill → Ongoing engagement
        ↓
Content Atomizer Skill → Email → social bridge

Stack combinations by business type

Use cases

Info / education

Courses, coaching, communities, digital products.

Challenges: selling transformation, building authority before the ask, skepticism toward “gurus,” high-ticket trust.

Angles that work: anti-guru, methodology transfer, community, results-first with specific people and timeframes.

Consulting / agency

Services, client work, B2B.

Challenges: limited inventory of time, crowded markets, “why you,” balancing delivery with acquisition.

Angles that work: specialist niche, named methodology, concrete client outcomes, anti-agency clarity.

Best clients attract best clients: content → reputation → referrals → case studies → more content.

E-commerce

Physical products, DTC, retail.

Challenges: competing beyond price, differentiation in crowded categories, email as revenue.

Angles that work: quality story, founder story, anti-mass-market, lifestyle identity.

Curate what customers already create—photos, reviews, unboxing—as marketing fuel.

SaaS

Software, apps, tools.

Challenges: feature parity, free/freemium pressure, trial-to-paid, churn.

Angles that work: simplicity, speed to value, integration with tools teams already use, anti-enterprise friction.

Every integration is a distribution channel—connecting tools connects audiences.

SEO: the long game

AI-assisted, human-verified. Quality over volume—find gaps, not the crowd.

• Research first with the Keyword Research skill
• Look for programmatic, scalable patterns
• One excellent piece beats five mediocre ones
• Add human checkpoints: expertise, fact-checking, claims
• Stagger publishing—velocity signals matter

Paid: creative at scale

Research → brief → generate → select winners. AI proposes variants; you choose what to test.

• Research before creating (Creative Strategist on competitor ads)
• Five to ten variants per concept; let testing pick
• Connect creative tooling (e.g. Glif MCP) for production
• You select two to three winners; platforms optimize delivery

Organic: the compound engine

One piece becomes many. Every asset compounds.

Content Atomizer workflow

INPUT: 1 Core Piece (blog, podcast, video, client work)
        ↓
Content Atomizer Skill
        ↓
OUTPUT: 15+ Pieces
├── LinkedIn: long post, carousel, poll
├── Twitter/X: thread, tweet, quote angle
├── Instagram: carousel, Reel script, stories
├── Email: newsletter section, nurture content
└── Video: YouTube script, Shorts script

Traffic flywheel

CREATE → ATOMIZE → VISUALIZE → DISTRIBUTE → CAPTURE → CONVERT → LEARN → REPEAT

Example outputs

Demo project: Boring Business Marketing—AI marketing for overlooked industries.

Landing page

Built in one session using Brand Voice, Positioning Angles, and Direct Response Copy. Research → shipped page.

boringbusinessmarketing.com — includes Perplexity research, “Confidently Boring” voice, “Overlooked Champion” angle, Direct Response copy, Frontend Design for HTML/CSS.

Video ad

Programmatic Remotion ad: terminal-style animation, 1920×1080, 16s @ 30fps—typing, stats, book reveal, CTA.

Sample: brand voice summary

Sounds like a smart friend who shares what actually works—without hype. Confidently boring: the unglamorous stuff that compounds.

Traits: practitioner not preacher, accessible expert, quietly contrarian. Use: “boring,” “compound,” specific numbers, “here's how.” Avoid: “revolutionary,” “10x,” empty buzzwords.

Sample: positioning angles (abbrev.)

• Overlooked Champion — AI marketing for industries “cool” agencies ignore.
• Compound Engine — the boring stuff that compounds while you sleep.
• Speed Advantage — what agencies do in months, ship in weeks.

All prompts

Copy, adapt, and run these to invoke the skills.

We are creating a [TYPE OF BUSINESS] targeting [AUDIENCE].
What makes us different: [UNIQUE APPROACH]

Use the Perplexity MCP to deeply research:
- Market landscape & key competitors
- Customer pain points & positioning gaps
- Pricing models & service packaging
- Best practices to repackage for our target

Save findings to /research folder as .md files.

Use the Brand Voice skill to analyze this content and
extract a voice profile:
[PASTE EXISTING CONTENT]

Output: voice-profile.md with traits, vocabulary guide,
do's and don'ts.

Use the Positioning Angles skill.

Context:
- Product: [WHAT YOU SELL]
- Transformation: [WHAT LIFE LOOKS LIKE AFTER]
- Competitors: [WHO THEY'D BUY FROM OTHERWISE]

Output: 3-5 angles with psychology and headline directions.

Use the Direct Response Copy skill.

Positioning angle: [CHOSEN ANGLE]
Brand voice: Reference voice-profile.md
Target audience: [SPECIFIC AUDIENCE]

Write landing page copy with:
- 3 headline options
- Hero section
- Problem/Solution sections
- Social proof sections
- CTA sections

Use the Email Sequences skill.

Trigger: [LEAD MAGNET NAME] download
Goal: [CONVERSION GOAL]
Brand voice: Reference voice-profile.md

Write 7-email welcome sequence following:
DELIVER → CONNECT → VALUE → VALUE → BRIDGE → SOFT → DIRECT

Spin up task-based agents to review this:
1. Agency Growth Expert
2. SEO Specialist
3. Conversion Expert
4. [Industry-specific expert]

Each analyzes from their perspective.
Synthesize: Where do they agree? What needs work?

Quick start checklist

Foundation setup

• Install VS Code
• Install Claude Code CLI and authenticate
• Configure Perplexity, Playwright, Glif MCPs as needed
• Download the Skills Pack
• Test: “Hey Claude, do you have access to the marketing skills?”

Before each session

• Create a project folder and a clear goal
• Gather brand assets and reference examples

During each session

• Start with research (thirty to sixty minutes minimum)
• Save context as markdown
• Checkpoint with expert review after major outputs
• Iterate—first drafts are starting points

After each session

• Summarize what shipped, note content opportunities, plan next session
• Ship something—even if imperfect