AI Ad System for Performance Marketing Agency

Project Summary

Media Made Simple is a performance-driven marketing agency specializing in e-commerce brands in crowded markets. We built an AI system that analyzes their Meta ad campaigns, identifies what makes their best ads work, and generates new creative concepts based on proven patterns.

The results: 50% reduction in creative production time, allowing their team to focus on strategy while scaling ad output without sacrificing quality.

How It Works

The system runs as an automated workflow that agency team members trigger through a simple form. They input a campaign ID and time window, then the AI handles the rest—analyzing ads, extracting insights, and generating new creative concepts. Results appear in their Airtable dashboard within 30 minutes to 2 hours, depending on campaign size.

The Challenge

Performance marketing agencies face constant pressure: clients need fresh ad creatives to test, but creating quality ads is time-intensive. Media Made Simple's team was spending hours analyzing campaign data and brainstorming new concepts, leaving less time for strategy and client relationships.

They needed a system that could handle the analytical heavy lifting while maintaining their brand-first creative approach.

Our Solution

We built a multi-stage AI automation using n8n that processes ads through three main phases:‍

Phase 1: Data Collection - The system connects to Facebook Graph API and pulls complete campaign data—all ads with their performance metrics (ROAS, CTR, CPM, conversions, engagement). Everything gets organized in Airtable for structured analysis.

Phase 2: Deep Analysis - This is where the AI does the heavy lifting. For each ad, the system:

• Transcribes video content to extract messaging and hooks
• Analyzes images to identify visual patterns and elements
• Breaks down ad copy using a standardized framework tracking psychological triggers and messaging angles
• Connects creative elements to performance metrics

Then it synthesizes insights across all ads, identifying which hooks, visuals, and messaging frameworks consistently drive results for this specific audience.

Phase 3: Creative Generation - Based on extracted insights, the AI generates detailed briefs for new ad concepts. Each brief includes recommended hooks, visual directions, copy frameworks, and calls-to-action—all grounded in what's already proven to work for this brand and audience.

The Technical Architecture

Built on n8n's automation platform with modular API integrations:

• Facebook Graph API for campaign data access
• Airtable & Supabase for structured data storage and workflow tracking
• Multiple AI providers (OpenAI, Anthropic, Google Gemini) for diverse analysis capabilities
• Scrapfly for additional web data when needed
• Google Workspace (Docs, Drive) for output management

The multi-provider AI approach ensures reliability—if one API is slow or down, the system adapts. The architecture is cost-optimized, leveraging free tiers where possible while maintaining enterprise-grade reliability.

The Interface

The system is designed for non-technical users. Team members access it through a simple web form where they input campaign details and click run. Progress updates appear in real-time in their Airtable dashboard, and final outputs—insights and new creative concepts—populate automatically when complete.

The Results

Media Made Simple cut creative production time in half. Their team now spends less time on data analysis and more time on strategic thinking and client relationships. They're testing more ad variations, moving faster, and maintaining the quality their clients expect—all without hiring additional team members.