How we Automated Meta Ads with 3 AI Systems

AI-Powered Meta Ad Systems for Agencies & Businesses

Key Concepts:

• AI Meta Ad Systems: Automated solutions leveraging AI (specifically Large Language Models - LLMs) to streamline Meta ad management and optimization.
• Airtable: A low-code platform used as a central database and front-end interface for these systems.
• N8N: A workflow automation tool used to connect different APIs and execute automated tasks.
• Replicate: A platform hosting various AI models (image & video) accessible via API.
• Perplexity AI: A search engine utilizing AI to provide comprehensive research results.
• Facebook Graph API: An API allowing access to Meta ad data for reporting and analysis.
• LLM (Large Language Model): AI models used for text generation, analysis, and prompt creation.
• Webhook: A method for one application to provide real-time information to another when an event occurs.

1. Automatic Product Ad Generator

This system addresses the time-consuming task of creating numerous product ads, particularly common for e-commerce businesses and agencies managing multiple clients. The core function is to automate ad creation based on product information and optional reference imagery.

Process:

1. Product Input: A new product is added to an Airtable workboard, including an image, product name, and notes.
2. Product Research (Optional): A Perplexity AI search is triggered to gather additional product details, supplementing the provided notes. This step can be manually adjusted.
3. Reference Image (Optional): A competitor ad or desired style image can be uploaded as a reference.
4. Ad Prompt Generation: Based on the product information and reference image (if provided), an LLM generates prompts for five ad variations.
5. Ad Image Generation: Using Nano Banana Pro via Replicate, five ad images are generated based on the prompts and the original product image. The aspect ratio and angle can be adjusted before generation.
6. Ad Copy Generation: Ad copy is automatically generated using the product information and research data.
7. Ad Gallery: All generated ads are displayed in an Airtable gallery.

Technical Details:

• Backend Workflows: Three primary workflows in N8N:
  • Workflow 1: Product research using Perplexity AI.
  • Workflow 2: Prompt creation (triggered when no reference image is uploaded), downloading the product image for context.
  • Workflow 3: Image generation using Nano Banana Pro via Replicate, looping through five variations.
• Reference Image Workflows: Two additional workflows mirror the above, triggered when a reference image is uploaded, incorporating both images into the prompt generation process.
• Cost: Generation costs are minimal (a few cents per ad).

Implementation: Requires cloning the Airtable template, connecting Airtable accounts, setting up webhooks, and providing API keys for Perplexity AI, the LLM, and Replicate.

2. Meta Ad Spying Tool

This system automates competitor ad analysis, providing weekly reports on competitor strategies. It aims to reduce the manual effort involved in tracking competitor advertising activities.

Process:

1. Competitor Input: Competitor URLs (specifically, the Facebook Ad Library link) are added to Airtable.
2. Ad Scraping: An Amplify Scraper scrapes the Facebook Ad Library pages of the listed competitors.
3. Data Formatting & Storage: Scraped data is formatted and stored in Airtable.
4. Ad Analysis: An LLM analyzes each new ad, providing insights and improvement suggestions.
5. Weekly Report Generation: An LLM analyzes all ads from a specific competitor and generates a weekly report summarizing their advertising strategy.

Technical Details:

• Workflow: A simpler workflow in N8N compared to the ad generator.
• Scraping: Uses Amplify Scraper to extract data from Facebook Ad Library pages.
• API Keys: Requires an Amplify API key and LLM API keys.

Implementation: Cloning the Airtable template, connecting Airtable, setting up the Amplify API key, and providing LLM API keys.

3. Automatic Meta Ad Campaign Reporting System

This system automates the creation of campaign performance reports, a significant time saver for agencies and businesses. It leverages data from Meta Ads, Google Sheets, and Google Slides.

Process:

1. Data Extraction: Data is extracted from Meta Ads using the Facebook Graph API.
2. Data Processing: Data is processed and organized in Google Sheets.
3. Report Generation: Google Slides are automatically populated with graphs, visuals, and conclusions based on the data in Google Sheets. The system compares campaigns, platforms, placements, and demographics.

Technical Details:

• Integration: Integrates N8N, Google Sheets, and Google Slides.
• API Connections: Requires connecting the Facebook Graph API, Google Slides API, and Google Sheets API.
• Google Zero Off: Setting up Google Zero Off is necessary for API access.

Implementation: Cloning the template, connecting the necessary APIs (Facebook Graph API, Google Slides API, Google Sheets API), and potentially customizing the template's branding.

Logical Connections & Overall Synthesis

The three systems are interconnected, forming a comprehensive suite for Meta ad management. The Ad Spying Tool provides inspiration and reference material for the Automatic Product Ad Generator. The Automatic Ad Campaign Reporting System then tracks the performance of the ads generated, completing the cycle.

The common thread across all three systems is the use of Airtable as a central database and N8N for workflow automation, coupled with the power of LLMs and platforms like Replicate for content creation. These systems are designed to reduce manual effort, improve efficiency, and provide data-driven insights for Meta advertising.

Notable Quote:

“Templates are great, but generally need to be customized a bit to each specific use case or client or for yourself.” – Speaker, emphasizing the need for adaptation and personalization.

Conclusion:

These AI-powered systems represent a significant opportunity for marketing agencies and businesses to automate and optimize their Meta advertising efforts. While implementation requires some technical setup, the potential time savings and improved performance justify the investment. The speaker highlights the availability of support and resources (the AI Accelerator) to assist with customization and implementation.