How to Build a Marketing Plan with AI (5-Day Process)

You’ve got your business idea. Maybe you’ve even built something.

Now what?

Most first-time founders skip this part. They go straight to “I need traffic” without knowing who they’re targeting or why those people would care.

Then they wonder why their ads don’t convert.

I’ve built marketing plans for dozens of businesses. The process used to take weeks of research, competitor analysis, and guesswork. Now I do it in a day using Claude.

This guide gives you the exact prompts I use. Copy them, customize for your business, and you’ll have a real marketing plan by tonight.

One warning: AI gives you hypotheses, not answers. The prompts below will get you 80% of the way there. The market validates the other 20% when you run actual campaigns.

Quick Navigation

Here’s what I’ve learned building businesses with AI:

Research used to be the bottleneck. Finding market data, analyzing competitors, understanding customers. It took weeks and most of it was guesswork anyway.

AI eliminates the bottleneck.

What used to take weeks now takes hours. The exact prompts below have helped me validate (and kill) business ideas before wasting money on ads.

The catch: most people use AI wrong. They ask “help me with my marketing” and get generic garbage back.

The fix: specific prompts with specific context. That’s what this guide gives you.

Don’t Make This Mistake

When I launched my first business, I targeted “small business owners who need marketing help.”

That’s 30 million people in the US alone. I was competing with agencies, freelancers, and every marketing tool ever built.

I burned through $3,000 in ads before I figured out the problem: I was targeting everyone, which meant I was targeting no one.

The fix was counterintuitive. I narrowed my target to “solo consultants making $100-200K who want to productize their services.”

That’s maybe 50,000 people. And suddenly my ads converted.

Peter Thiel explains this in Zero to One. When PayPal launched, they didn’t target “online shoppers.” They targeted eBay PowerSellers - about 20,000 people who desperately needed easier payments.

PayPal dominated that niche, then expanded.

This is your template. If you’re starting a business:

1. Pick the narrowest viable audience
2. Become the obvious choice for those people
3. Expand only after you’ve dominated

The prompts below help you find that narrow target. Don’t skip this step. Every failed campaign I’ve run traces back to targeting too broadly.

Step 1: Market Research with AI

Before I build anything, I run market research. This used to take days of reading industry reports and analyst opinions. Now it takes 30 minutes.

Here’s what you need to know before spending money:

• How big is this market, and is it growing?
• Who are the main players and how are they positioned?
• What do customers actually complain about?
• Where are the gaps?

The prompts below get you answers fast. But remember: AI pulls from training data, which can be outdated. Use these outputs as a starting point, then validate with real customer conversations when possible.

Market Overview Prompt

Open Claude or ChatGPT and paste this. Replace the brackets with your specifics:

You are a market research analyst. Give me a comprehensive overview
of the [INDUSTRY/NICHE] market.

Include:
1. Estimated market size and growth trajectory
2. Key players and their approximate market share
3. Main customer segments and their characteristics
4. Current trends and disruptions affecting the market
5. Common pain points customers face
6. Typical price points and business models
7. Barriers to entry for new competitors
8. Regulatory or compliance considerations

Be specific with numbers where possible. Cite your reasoning.

Industry: [YOUR INDUSTRY]
Geographic focus: [US/Global/Specific region]

Real example: When I was evaluating a productized SEO audit service, I used “technical SEO services for SaaS companies” as my industry. The output told me the market was $2-3B, dominated by agencies charging $5-15K/month, with a gap for one-time audits under $1,000.

That gap became my positioning.

Once you have the overview, run this follow-up:

Based on this overview, what are the 3 most underserved
customer segments in this market? For each segment, explain:
- Why they're underserved
- What they're currently doing to solve the problem
- What would make them switch to a better solution

This gives you your first list of potential targets. Save both outputs - you’ll use them in the next step.

Step 2: Building Buyer Personas with AI

Generic personas are useless. “35-year-old marketing manager who wants to grow their business” describes millions of people and helps you target none of them.

You need personas specific enough that you could find them on LinkedIn. Specific enough that you know which Facebook groups they’re in. Specific enough that you know the exact words they use to describe their problem.

The trick: give AI your market research from Step 1, then ask for multiple distinct personas to compare.

Buyer Persona Prompt

You are a customer research specialist. Based on the following
product/service, create 3 distinct buyer personas.

Product/Service: [DESCRIBE WHAT YOU'RE SELLING]
Price Point: [APPROXIMATE PRICE]
Main Benefit: [PRIMARY VALUE PROPOSITION]
Industry Context: [PASTE YOUR MARKET OVERVIEW FROM STEP 1]

For each persona, provide:

DEMOGRAPHICS:
- Age range
- Gender (if relevant)
- Income level
- Job title/role
- Location characteristics

PSYCHOGRAPHICS:
- Values and priorities
- Buying decision factors (price vs. quality vs. speed vs. status)
- Where they spend time online
- What content they consume
- Who influences their decisions

PAIN POINTS:
- Primary problem your product solves for them
- Secondary problems
- What they've tried before that didn't work
- What's stopping them from solving this problem

OBJECTIONS:
- Why they might NOT buy
- What would make them hesitant
- What competitors they'd consider instead

BUYING TRIGGERS:
- What event or realization would make them search for a solution
- What would make them buy TODAY vs. "someday"

Make each persona distinctly different. Give them names and make
them feel like real people.

The key: You want personas different enough that you could target them separately. If your three personas all sound similar, you haven’t gone deep enough.

Persona Validation Prompt

Now pick your winner. This is the most important decision you’ll make:

Looking at these 3 personas, answer:

1. Which persona has the most urgent need? Why?
2. Which persona has the highest willingness to pay? Why?
3. Which persona would be easiest to reach with paid advertising?
   What platforms and targeting would work?
4. Which persona has the shortest sales cycle?
5. If I could only target ONE persona to start, which should it
   be and why?

The answer to question 5 becomes your primary target. Ignore the other two for now.

I know it feels like you’re leaving money on the table. You’re not. Trying to target all three means your messaging speaks to no one. Start with one, nail it, then expand.

Step 3: Competitive Analysis with AI

Your competitors have already spent money figuring out what works. Study them.

I analyze competitors before every campaign. Not to copy them, but to find the gaps they’re missing. The positioning no one owns. The pain points no one addresses.

Method 1: Website Copy Analysis

Here’s my process:

1. Google your main keyword (what you’d want to rank for)
2. Open the top 5 organic results and top 3 ads
3. Copy the homepage and main sales page text from each
4. Paste it all into Claude with this prompt:

Analyze these competitor websites. For each one, identify:

1. VALUE PROPOSITION: What's their main promise? Who are they targeting?
2. POSITIONING: Are they premium, budget, or mid-market?
   What makes them different?
3. OBJECTIONS ADDRESSED: What concerns do they preemptively handle?
4. SOCIAL PROOF: What credibility signals do they use?
5. CTA STRATEGY: What action do they want visitors to take?
6. GAPS: What pain points or objections are NOT addressed?

[PASTE COMPETITOR 1 COPY]

---

[PASTE COMPETITOR 2 COPY]

---

[Continue for each competitor]

Method 2: Ad Copy Analysis

If competitors are running ads, they’re telling you what works. Use a spy tool like AdPlexity to pull their ads, or search Facebook Ad Library for free.

Paste the ads into Claude:

Analyze these competitor advertisements. Identify:

1. ANGLES: What hooks or angles are they using?
2. TARGETING SIGNALS: Based on the messaging, who are they targeting?
3. FUNNEL STAGE: Are these ads targeting cold, warm, or hot traffic?
4. PATTERNS: What do multiple competitors do similarly?
5. GAPS: What angles or audiences are NO ONE targeting?

[PASTE AD COPY AND DESCRIPTIONS]

Competitive Gap Analysis Prompt

This is where it gets good. You’re looking for what no one else is doing:

Based on this competitive analysis, identify:

1. POSITIONING GAPS: What positioning is no competitor owning?
2. AUDIENCE GAPS: What customer segments are underserved?
3. MESSAGING GAPS: What pain points are competitors ignoring?
4. CHANNEL GAPS: What traffic sources are competitors not using?
5. OFFER GAPS: What offers or guarantees could differentiate us?

For each gap, rate:
- Opportunity size (1-10)
- Difficulty to execute (1-10)
- Fit with our product (1-10)

What you’re looking for: A gap with high opportunity, low difficulty, and good fit. That’s your positioning.

When I analyzed the SEO audit market, I found everyone was selling ongoing monthly retainers. The gap was one-time audits with implementation guides. That became my offer.

Step 4: Opportunity Identification

You’ve done the research. Now you pick your lane.

At this point you have:

• Market overview with size and trends
• One primary persona (your ideal customer)
• Competitive analysis showing how others are positioned
• Gap analysis showing what’s missing

Time to synthesize this into a specific opportunity you can test.

Opportunity Synthesis Prompt

Based on all the research below, identify the top 3 market
opportunities for [YOUR PRODUCT].

For each opportunity, provide:

1. TARGET: Which persona or segment
2. POSITIONING: How we'd position against competitors
3. ANGLE: The main hook or message
4. CHANNEL: Best traffic sources to reach them
5. OFFER: What we'd offer to get them to act
6. VALIDATION: How we'd test this before scaling

Rank opportunities by:
- Revenue potential (1-10)
- Speed to validate (1-10)
- Competitive defensibility (1-10)

[PASTE YOUR RESEARCH FROM STEPS 1-3]

Pick the opportunity with the highest “speed to validate” score. Not the biggest revenue potential. Not the most defensible.

Why? Because you need to learn fast. Your first hypothesis is probably wrong. You want to find out quickly and iterate.

Revenue potential matters later, once you’ve validated that people actually want what you’re selling.

Step 5: Funnel Strategy

You know who you’re targeting and how you’re positioned. Now you need a path from “stranger who’s never heard of you” to “paying customer.”

This isn’t complicated, but most people overthink it.

Most New Businesses Make This Mistake

They run ads saying “Buy Now” to people who’ve never heard of them.

That works for Amazon. It doesn’t work for your new business.

You’re competing with established players for the 3% of people ready to buy today. And you’re losing, because they have more trust and bigger budgets.

The smarter play: Target the 97% who have the problem but aren’t ready to buy yet. They’re cheaper to reach and less competitive.

Top-of-Funnel Strategy

Capture people before they’re ready to buy. Then nurture them until they are.

Real example from a client selling project management software:

• Bottom of funnel ad: “Try [Software] Free - Project Management Made Easy”
• Top of funnel ad: “Why Your Projects Keep Going Over Budget (Free Audit Template)”

The bottom ad got a 0.8% CTR and $12 CPL. The top ad got a 2.4% CTR and $3 CPL.

The top-of-funnel leads took longer to convert. But they converted at a higher rate because they’d been nurtured, and the overall CAC was lower.

The trade-off is real: Top-of-funnel leads need nurturing. You need email sequences, retargeting, content. It’s more work upfront.

But for a new business without brand recognition, it’s often the only way to compete.

Lead Magnet Strategy with AI

Your lead magnet is the bridge between “stranger” and “subscriber.” It needs to be good enough that people give you their email for it.

Create 5 lead magnet ideas for [PRODUCT/SERVICE].

Target persona: [PASTE YOUR PRIMARY PERSONA]
Main pain point: [THE PROBLEM YOU SOLVE]

For each lead magnet idea, provide:
1. Title (benefit-driven, specific)
2. Format (PDF, video, quiz, calculator, etc.)
3. What it teaches or provides
4. Why this persona would want it
5. How it naturally leads to your paid product

Make them progressively different - don't just give me 5 variations
of the same thing.

Pick the one that:

• Solves a small piece of their main problem
• Can be consumed in under 10 minutes
• Naturally leads them to want your paid solution

I’ve found calculators and templates convert better than PDFs. People actually use them, which builds trust.

Lead Nurturing Sequence with AI

Most people capture leads and then immediately pitch. That’s like proposing on a first date.

Instead, run a nurture sequence that builds trust over 2 weeks:

Create a 5-email nurturing sequence for leads who downloaded
[LEAD MAGNET NAME].

Goal: Move them from problem-aware to solution-aware to
product-aware over 2 weeks.

Email 1 (Day 0): Deliver the lead magnet + quick win
Email 2 (Day 2): Expand on the problem, share a story
Email 3 (Day 5): Introduce the solution category (not your product yet)
Email 4 (Day 8): Show proof that the solution works
Email 5 (Day 12): Introduce your product, soft CTA

For each email, provide:
- Subject line (3 options)
- Email body (300-400 words)
- Call to action

Tone: Direct, helpful, not salesy. Write like a friend who
happens to be an expert.

This sequence does the heavy lifting. By email 5, they know you, trust you, and understand why your solution makes sense.

Set up your email automation →

Step 6: Traffic Source Planning

Don’t try to be everywhere. Pick one or two channels to start, based on where your persona actually spends time.

For most new businesses, I recommend starting with either Google Search (if people search for your solution) or Facebook (if they don’t).

Google is higher intent but more competitive. Facebook is cheaper but requires better creative and nurturing.

Traffic Source Selection Prompt

Let AI recommend based on your specific situation:

Based on this buyer persona and marketing strategy, recommend
the top 2 traffic sources to test first.

Persona: [PASTE YOUR PRIMARY PERSONA]
Product: [YOUR PRODUCT]
Price point: [PRICE]
Funnel strategy: [TOP/MIDDLE/BOTTOM FOCUS]
Budget for testing: [MONTHLY BUDGET]

For each traffic source, provide:
1. Why it fits this persona
2. Targeting strategy (interests, keywords, audiences)
3. Ad format recommendations
4. Expected CPM/CPC range
5. Minimum test budget needed
6. What "success" looks like for a test

Rank by: Likelihood of profitable results within 30 days

Important: I said “top 2” not “top 3.” You don’t have the budget or attention to test three channels well. Pick two, run them properly, then expand.

Keyword Research with AI

If you’re running Google Ads, you need keywords. This prompt gets you started:

Generate keyword ideas for [PRODUCT/SERVICE].

Include:
1. HIGH INTENT (buyer keywords): People ready to purchase
2. COMPARISON (consideration stage): People evaluating options
3. PROBLEM-AWARE: People searching for solutions to the problem
4. INFORMATIONAL: People researching the topic

For each keyword, estimate:
- Search volume (high/medium/low)
- Competition (high/medium/low)
- Buyer intent (1-10)
- Recommended landing page type

Focus on long-tail keywords with lower competition that
bigger competitors might miss.

Start with high-intent keywords. Yes, they’re more competitive. But people searching “best [solution] for [specific use case]” are ready to buy. Lower-intent keywords require more nurturing.

Full Google Ads setup guide →

The Complete AI Marketing Plan Workflow

Here’s my 5-day process for going from zero to launching your first test campaign:

Day 1: Research Foundation

1. Run the market overview prompt - save the output
2. Generate 3 buyer personas using the market overview
3. Pick your primary persona using the validation prompt
4. Write down: “I’m targeting [specific persona] because [reason]”

Day 2: Competitive Intelligence

1. Google your main keyword, open top 8 results
2. Copy competitor website text and ads
3. Run competitive analysis prompts
4. Identify your positioning gap
5. Write down: “I’m positioned as [X] because competitors ignore [Y]”

Day 3: Funnel & Offer Strategy

1. Decide: top-of-funnel (lead magnet) or bottom-of-funnel (direct offer)
2. If lead magnet: generate ideas, pick one, create it with AI
3. Map out your email sequence
4. Select 2 traffic sources to test

Day 4: Campaign Assets

1. Write landing page copy with AI (then edit heavily)
2. Generate 10+ ad variations per angle
3. Build your email sequence in ConvertKit or similar
4. Create simple retargeting for visitors who don’t convert

Day 5: Launch

1. Review everything - AI output needs human editing
2. Set up tracking: pixels, conversion events, UTM parameters
3. Build campaigns in your ad platforms
4. Launch with $20-50/day per channel
5. Set calendar reminder to review in 7 days

Full tracking setup guide →

What AI Can and Can’t Do

I want to be honest about limitations.

AI is excellent at:

• Compressing research from weeks to hours
• Generating hypotheses and angles to test
• Creating first drafts of copy and content
• Identifying patterns in competitor data
• Structuring your thinking

AI cannot:

• Tell you if people will actually buy
• Validate your price point
• Guarantee your ads will convert
• Replace testing with real traffic and real money

The market decides what works. AI just helps you build better hypotheses faster.

Your real advantage: While competitors spend weeks on research, you’ve already launched and gathered real data. You’re on iteration 3 while they’re still planning iteration 1.

Advanced: Custom AI Workflows for Marketing

Everything above works by copying prompts into ChatGPT or Claude.

But if you’re running an AI-powered business, you’ll want to automate this. Run your marketing research on autopilot while you focus on building.

SKILLS.md Files (Claude Code)

If you’re using Claude Code, you can save your methodology as reusable skills.

What this means: Instead of copying the same prompts over and over, you define them once. Every time you ask Claude to “research this market,” it follows YOUR process automatically.

Example: Marketing Research Skill

Create a file called market-research.md:

# Market Research Skill

When I ask you to research a market, follow this exact process:

## Step 1: Market Overview
- Market size and growth
- Key players and positioning
- Customer segments
- Current trends and disruptions

## Step 2: Competitive Analysis
For each competitor I mention:
- Value proposition
- Target audience
- Pricing model
- Strengths and weaknesses
- Gaps in their offering

## Step 3: Opportunity Identification
Based on the above, identify:
- Underserved segments
- Unaddressed pain points
- Positioning opportunities
- Potential angles to test

## Output Format
Always structure your response with clear headers.
Include a "Quick Wins" section at the end with
3 actionable opportunities I can test this week.

Now when you ask Claude to “research the [X] market,” it follows YOUR methodology every time.

Skills I use for my own marketing:

The real power: Chain skills together. “Use my market-research skill for the fitness coaching market, then use persona-builder to create 3 personas, then use ad-copy-generator to write 10 ads for each persona.”

One command, 30+ usable ad variations.

MCP Servers (Real-Time Data)

This is where AI goes from “helpful assistant” to “actual business automation.”

MCP (Model Context Protocol) servers connect Claude to your live data:

• Google Analytics
• Your ad platform stats
• CRM and customer data
• Competitor websites (monitoring changes)
• Your database

Example: Instead of exporting data and pasting it into Claude, you say:

"Pull my Google Analytics data for the last 30 days. Find pages
with high traffic but low conversion rates. Suggest 3 improvements
for each underperforming page."

Claude pulls the data, analyzes it, and gives you recommendations. No manual export required.

MCP servers I use:

Setting up MCP requires technical skills. You’ll need:

1. Claude Code installed
2. Node.js on your machine
3. The specific MCP server packages

If you’re not technical, hire someone on Upwork to set it up once. Or use the basic copy-paste workflow above - it works fine for most businesses.

MCP server setup guide →

Where to start:

Your Marketing Plan Template

After running the prompts above, fill this out. One page, nothing fancy:

MARKETING PLAN: [PRODUCT/BUSINESS NAME]
Date: [TODAY]

WHO I'M TARGETING
- Primary Persona: [Name, job title, specific situation]
- Their Main Pain Point: [The problem they're trying to solve]
- Why They'd Care About Me: [What makes my solution relevant]

MY POSITIONING
- Competitive Gap I'm Filling: [What no one else is doing]
- My Angle: [How I'm different]
- Price Position: [Premium/Mid/Budget]

MY FUNNEL
- Traffic Source 1: [Platform + targeting approach]
- Traffic Source 2: [Platform + targeting approach]
- Lead Magnet: [What I'm offering for their email]
- Primary Offer: [What I'm selling]
- Price: [$ amount]

SUCCESS METRICS (30-day targets)
- Cost Per Lead: $[X]
- Lead to Customer Rate: [X]%
- Customer Acquisition Cost: $[X]
- Revenue Goal: $[X]

FIRST 3 TESTS (This week)
1. [Specific test with measurable outcome]
2. [Specific test with measurable outcome]
3. [Specific test with measurable outcome]

Print this. Put it where you can see it. Update it after every test.

Your Next Step

You have two options:

Option 1: Do it yourself

1. Block 2 hours tomorrow morning
2. Open Claude or ChatGPT
3. Run through the prompts in this guide
4. Fill out your one-page marketing plan
5. Launch your first test within 7 days

Don’t spend weeks perfecting. Spend days building, then let real data tell you what to fix. The best marketing plan is the one that’s tested, not theorized.

Option 2: Do it with me Book a strategy call and we’ll build your marketing plan together in 90 minutes. You’ll leave with a completed plan, your first ad variations, and a clear path to launch.

Related Guides

These fit into the Plan → Build → Acquire → Optimize framework:

Planning:

• Niche Research with AI - Validate your idea before building
• Competitor Analysis - Deep dive on what competitors are doing

Building:

• Landing Page Creation - Build pages that convert
• Content Site Guide - If content is part of your funnel

Acquiring Traffic:

• Google Ads Guide - Search traffic for high-intent buyers
• Facebook Ads Guide - Social traffic for awareness

Optimizing:

• Campaign Optimization - Improve ROI once you launch
• AI Tools for Marketing - Full AI workflow and prompts