How to Build a Dynamic Ad Copy Personalization System

Dynamic ad copy personalization automation generates different copy for different audience segments without you writing each version by hand. I use this to produce 20+ variations from a single core message, each tailored to a specific segment's language, pain points, and motivation.

The difference between generic and personalized copy is often 30-50% in CTR. Same offer, same page, different words.

What You Need Before Starting

• Anthropic API key
• Python 3.8+ with anthropic installed
• Audience segment definitions with pain points
• A core offer and value proposition

Step 1: Define Your Audience Segments

SEGMENTS = [
    {
        "name": "Agency Owners",
        "pain": "Managing 10+ ad accounts manually, team cannot keep up",
        "language": "Talks about clients, retainers, margins, team overhead",
        "motivation": "More profit per client, fewer hours working IN the business",
        "objection": "Already tried AI tools, they were not reliable",
    },
    {
        "name": "Solo Consultants",
        "pain": "Doing everything themselves, no time to scale",
        "language": "Talks about bandwidth, wearing too many hats, can't take more clients",
        "motivation": "Take on more clients without hiring, automate the repetitive stuff",
        "objection": "Cannot afford to invest in systems right now",
    },
    {
        "name": "Ecommerce Operators",
        "pain": "Ad costs rising, ROAS declining, cannot find winning products fast enough",
        "language": "Talks about ROAS, AOV, margins, inventory, scaling",
        "motivation": "Better ROAS, faster testing, automated scaling decisions",
        "objection": "Their store is too unique for templates to work",
    },
]

Step 2: Generate Personalized Copy

import anthropic
import json

def generate_personalized_copy(core_offer, segment, framework="PAS"):
    client = anthropic.Anthropic()
    
    prompt = f"""Write Meta ad copy personalized for this specific audience segment.

CORE OFFER: {core_offer}
FRAMEWORK: {framework}

AUDIENCE SEGMENT:
Name: {segment['name']}
Pain: {segment['pain']}
Language: {segment['language']}
Motivation: {segment['motivation']}
Top objection: {segment['objection']}

RULES:
- Use their exact language patterns (not yours)
- Address their specific pain in the hook
- Handle their objection naturally in the body
- CTA matches their motivation
- Grade 5-6 reading level
- Use contractions
- Max 125 words primary text
- Max 40 character headline

Generate 3 variations with different hooks:
1. Pain-driven hook
2. Curiosity hook
3. Social proof hook

Return JSON: [{{"hook_type": "...", "primary_text": "...", "headline": "...", "description": "..."}}]"""

    resp = client.messages.create(
        model="claude-sonnet-4-20250514",
        max_tokens=1500,
        messages=[{"role": "user", "content": prompt}]
    )
    return json.loads(resp.content[0].text)

Step 3: Generate the Full Matrix

def generate_copy_matrix(core_offer, segments, frameworks=None):
    if not frameworks:
        frameworks = ["PAS", "BAB", "AIDA"]
    
    matrix = []
    for segment in segments:
        for framework in frameworks:
            copies = generate_personalized_copy(core_offer, segment, framework)
            for copy in copies:
                copy["segment"] = segment["name"]
                copy["framework"] = framework
                matrix.append(copy)
    
    print(f"Generated {len(matrix)} copy variations across {len(segments)} segments and {len(frameworks)} frameworks")
    return matrix

Step 4: Score for Personalization Quality

def score_personalization(copy_text, segment):
    client = anthropic.Anthropic()
    
    resp = client.messages.create(
        model="claude-sonnet-4-20250514",
        max_tokens=300,
        messages=[{"role": "user", "content": f"""Score this ad copy for personalization quality.

Target segment: {segment['name']}
Their language: {segment['language']}
Their pain: {segment['pain']}

Copy: {copy_text}

Score 1-10 on:
- Language match (does it use their words?)
- Pain specificity (does it name their exact problem?)
- Relevance (would they feel "this is for me"?)
- Objection handling (does it address their hesitation?)

Return JSON: {{"language": N, "pain": N, "relevance": N, "objection": N, "total": N, "fix": "..."}}"""}]
    )
    return json.loads(resp.content[0].text)

Step 5: Export for Campaign Creation

import csv

def export_copy_matrix(matrix, output_path="personalized_copy.csv"):
    with open(output_path, "w", newline="", encoding="utf-8") as f:
        writer = csv.DictWriter(f, fieldnames=[
            "segment", "framework", "hook_type", "primary_text", "headline", "description"
        ])
        writer.writeheader()
        for row in matrix:
            writer.writerow({
                "segment": row["segment"],
                "framework": row["framework"],
                "hook_type": row["hook_type"],
                "primary_text": row["primary_text"],
                "headline": row["headline"],
                "description": row.get("description", ""),
            })
    print(f"Exported {len(matrix)} variations to {output_path}")

Each segment gets its own ad set with copy that speaks directly to them. Same landing page, same offer, but the ad feels like it was written specifically for that person.

What to Build Next

Add performance tracking by segment to see which personalization angles convert best. Then build a feedback loop that identifies the top-performing segment language and feeds it back into the copy generator.

Related Reading

• Audience Research with AI - understanding your audience before personalizing
• AI for Creative Strategy and Testing - strategic creative testing
• The Data Flywheel Explained - how data improves personalization over time