What is Ad Variation Engine?

An AI tool that generates 10 ad variations from a single winning ad using PAS, AIDA, and BAB frameworks.

Where are the outputs stored?

The tool writes ad_test_matrix.csv in the project directory and creates winning_ad.txt if missing.

Can I customize frameworks or the number of variations?

Yes. You can adjust the target frameworks and increase or decrease the number of variations to test.

Ad Variation Engine

npx machina-cli add skill akhilkannur/marketing-agent-blueprints/ad-variation-engine --openclaw

Files (1)

SKILL.md

1.6 KB

The Ad Variation Engine

Core Instructions

You are a highly specialized AI agent focusing on Paid Media. Your mission is: Takes one winning ad (yours or a competitor's) and generates 10 high-quality variations using proven psychological frameworks like PAS, AIDA, and BAB.

Implementation Workflow

Phase 1: Initialization

Check: Does winning_ad.txt exist?
If Missing: Create it using sampleData.
Read: Load the core value proposition.

Phase 2: The Mutation Lab

For the input ad, generate the following variations:

Cluster 1: The Pain (PAS)

Variation A (Agitation): Focus heavily on the "Spreadsheet Hell."
Variation B (Fear): "Your competitors are automating. You are manual."

Cluster 2: The Logic (AIDA)

Variation C (Attention): "Sales Directors: Read this."
Variation D (Desire): "Imagine closing 20% more deals."

Cluster 3: The Pattern Interrupt (Social)

Variation E (Reviews): "Rated 4.9/5 by [Industry] Reps."
Variation F (Contrast): "Old way: Excel. New way: Us."

Phase 3: Output

Generate ad_test_matrix.csv:

Framework: [PAS / AIDA / Social]
Headline: [Hook]
Body_Copy: [Ad Text]
Hypothesis: "Testing if Fear drives cheaper clicks than Logic."

Blueprint ID: ad-variation-engine Source: Real AI Examples

Source

git clone https://github.com/akhilkannur/marketing-agent-blueprints/blob/main/skills/ad-variation-engine/SKILL.mdView on GitHub

Overview

Ad Variation Engine takes a single winning ad and generates 10 high quality variations using proven psychological frameworks like PAS, AIDA, and BAB. It creates multiple angles to boost CTR and conversions. The output is a structured ad_test_matrix.csv with framework, headline, body copy, and a testing hypothesis.

How This Skill Works

The tool checks for winning_ad.txt and creates it from sampleData if missing. It then reads the core value proposition and applies PAS, AIDA, and BAB patterns to craft ten variations. Finally it outputs ad_test_matrix.csv with fields Framework, Headline, Body_Copy, and Hypothesis for testing.

When to Use It

Launching a quick creative test by converting a winning ad into 10 variants across frameworks
Optimizing CTR and CVR by probing pain, logic, and social proof angles
Scaling paid campaigns with consistent structure ready for A/B tests
Comparing emotional vs logical messaging to identify cheaper clicks
Producing ready to test ad copy for new campaigns or platforms

Quick Start

Step 1: Ensure winning_ad.txt exists or create it from sampleData
Step 2: Run the Ad Variation Engine to generate 10 variations using PAS, AIDA, and BAB
Step 3: Open ad_test_matrix.csv to review and select variations for testing

Best Practices

Start from the core value proposition of the winning ad
Diversify across PAS, AIDA, and Social patterns to cover different intents
Keep headline length concise and align with the value prop
Include a clear testing hypothesis for each variation
Review brand voice and policy compliance before deployment

Example Use Cases

PAS variation focusing on Spreadsheet Hell that leads to calmer workflows
AIDA variation with Attention prompt Sales Directors read this
Social pattern variation with Reviews and social proof from industry reps
BAB variation showing Before After Bridge in a SaaS ad
ad_test_matrix.csv output example showing Framework, Headline, Body_Copy, Hypothesis

Frequently Asked Questions

Add this skill to your agents