Video Ad Analyzer

Analyze marketing videos and images using Google's Gemini AI to extract marketing dimensions, hooks, messaging patterns, and strategic insights.

What It Does

The analyzer processes video ads and extracts:

• Hooks: Opening hooks (visual, spoken, text overlays)
• Social Proof: Testimonials, reviews, authority signals
• Pain Points: Problems being addressed
• Benefits: Value propositions and outcomes
• CTAs: Calls to action and their placement
• Urgency: Scarcity and time pressure tactics
• Offers: Pricing, deals, bundles
• Objection Handling: How concerns are addressed
• Emotional Triggers: Emotional appeals used
• Visual Style: Design and aesthetic choices
• Messaging Sequence: How the message unfolds

Requirements

• Python 3.9+
• Google Cloud account with Vertex AI enabled
• Service account with Vertex AI permissions

Quick Start

1. Clone and Install

git clone https://github.com/YOUR_USERNAME/video-ad-analyzer.git
cd video-ad-analyzer
pip install -r requirements.txt

2. Set Up Google Cloud

1. Create a Google Cloud project at console.cloud.google.com

2. Enable the Vertex AI API:

   APIs & Services > Enable APIs > Search "Vertex AI" > Enable
   

3. Create a service account:

   IAM & Admin > Service Accounts > Create Service Account
   

   • Name: video-analyzer
   • Role: Vertex AI User

4. Download the JSON key:

   Service Account > Keys > Add Key > Create new key > JSON
   

3. Configure

cp config.example.py config.py

Edit config.py:

SERVICE_ACCOUNT_PATH = "/path/to/your-service-account.json"
PROJECT_ID = "your-project-id"

4. Run Analysis

# Analyze a single video
python analyze.py --video path/to/ad.mp4

# Analyze all videos in a directory
python analyze.py --directory path/to/ads/

# Limit batch size
python analyze.py --directory path/to/ads/ --max 10

Output

Results are saved to analysis_output/ (configurable) as JSON files:

analysis_output/
└── 2024-01-15_10-30-45/
    ├── consolidated.json      # All dimensions in one file
    ├── hooks.json
    ├── social_proof.json
    ├── pain_points.json
    ├── benefits.json
    ├── cta.json
    └── ...

Example Output

{
  "hooks": {
    "visual_hook": {
      "description": "Close-up of frustrated person looking at phone",
      "timestamp": "0:00-0:03",
      "effectiveness": "High - immediately relatable emotion"
    },
    "spoken_hook": {
      "text": "Tired of apps that don't actually work?",
      "tone": "Empathetic, slightly frustrated"
    },
    "text_hook": {
      "text": "STOP SCROLLING",
      "style": "Bold white text, red background"
    }
  }
}

Configuration Options

See config.example.py for all options:

Option	Description	Default
DEFAULT_MODEL	Gemini model to use	gemini-2.0-flash
API_CALL_DELAY	Seconds between API calls	30
NATIVE_ANALYSIS_TIMEOUT	Max time for video analysis	300
RUN_STRATEGIC_ANALYSIS	Run pattern aggregation	True

Supported Formats

Video: .mp4, .mov, .avi, .mkv, .webm

Image: .jpg, .jpeg, .png, .gif, .bmp, .webp

Cost Estimate

Using gemini-2.0-flash (recommended):

• ~$0.01-0.03 per video (depending on length)
• ~$0.005 per image

Troubleshooting

"Service account file not found"

Make sure SERVICE_ACCOUNT_PATH in config.py points to your downloaded JSON key file.

"Permission denied" errors

Ensure your service account has the Vertex AI User role.

Rate limit errors

Increase API_CALL_DELAY in config.py (default: 30 seconds).

Video processing fails

Install ffmpeg for audio extraction:

# macOS
brew install ffmpeg

# Ubuntu
sudo apt install ffmpeg

License

MIT License - Use freely in your projects.