How we Automated a Meta Ad Agency with AI [Free Template]

Meta Ad Agency Automation in Naden and Air Table: Detailed Summary

Key Concepts:

• Meta Ads Automation
• No-Code Automation
• Air Table Interface
• NAN Automation
• Ad Angle Generation
• Ad Script Generation
• VSSL (Video Sales Script) Structure
• Ad Performance Tracking
• Facebook Graph API
• Web Scraping (Scrapfly)
• Perplexity AI
• Google Gemini 2.5 Pro
• Return on Ad Spend (ROAS)

1. System Overview and Workflow

The video details a system built to automate a Meta ad agency's workflow using Air Table and NAN, enabling automated ad campaign tracking, ad idea generation, ad copy creation, and video script generation based on top-performing ads. The system replicates the traditional ad agency cycle:

• Research: Brand, audience, products/services.
• Ideation: Ad angles, ad copies, scripts, creative design.
• Monitoring: Daily performance analysis.
• Optimization: Improving existing/new ads based on performance data.

The system uses Air Table as a front-end interface for human input and decision-making, while NAN handles back-end automation and execution. The workflow is structured in five phases:

1. Initial Product Research: Adding new/existing products and conducting in-depth research.
2. Ad Angle/Idea Generation and Selection: Generating ad angles/ideas, with human selection.
3. Script Generation and Review: Generating scripts, iterating, and reviewing with AI and human input.
4. Ad Performance Tracking: Analyzing ad performance data.
5. Winning Script/Ad Selection: Using top-performing ads as a basis for new ads.

2. Air Table Interface and Functionality

The Air Table interface, built using Air Table Interfaces, presents a user-friendly way to manage the ad creation process. It's essentially a front-end for a standard Air Table database. The interface visualizes the five-step workflow.

• Phase 1 (Initial Product Research):
  • Adding products, brands, and notes.
  • Initiating deep research on product details, customer profiles, and brand alignment using Perplexity.
  • Incorporating winning ad scripts from past campaigns.
• Phase 2 (Ad Angle Selection):
  • Generating and displaying various ad angles with details like concept summary, VSSL structure (hook, problem, solution, proof, credibility, offer, CTA), video suggestions, emotional triggers, cognitive biases, and direct response techniques.
  • Shortlisting preferred angles.
• Phase 3 (Script Generation):
  • Selecting script length and generating ad scripts based on chosen angles.
  • Reviewing and revising scripts with feedback, triggering script regeneration.
• Phase 4 (Ad Performance Tracking):
  • Connecting to Meta ad accounts to track campaign performance.
  • Monitoring metrics like spend, ROAS (Return on Ad Spend), CPC (Cost Per Click), clicks, conversions, and impressions.
  • Selecting top-performing ads based on ROAS.
• Phase 5 (Winning Scripts):
  • Storing best-performing video scripts.
  • Associating winning scripts with new products to inform new ad creation.

3. NAN Automation Breakdown

The NAN automation is divided into three main parts:

1. Ad Performance Analytics:
   • Scheduled trigger (daily/weekly) to fetch data from Meta ad accounts.
   • Outputting performance data to Air Table (Phase 4).
2. Best Performing Video Ad Scraping:
   • Triggered by a web hook when a "best ad" is selected in Air Table (Phase 4).
   • Scrapes the video script of the best-performing ad.
3. Generation:
   • Four actions triggered by a single web hook:
     • Generate product details and customer research (triggered in Phase 1).
     • Generate ad angles.
     • Generate scripts.
     • Provide feedback for script iteration.

3.1. Generating Product Details and Customer Profile

• Triggered by a web hook when "Generate Product Details" is selected in Air Table.
• Uses a switch module to route the web hook to the correct automation flow.
• Retrieves product information from Air Table using the record ID.
• Constructs a prompt for Perplexity AI, including product name and notes.
• Uses Perplexity (Sonar Reasoning model) to conduct deep research.
• Updates the Air Table record with product details and customer ICP (Ideal Customer Profile).

3.2. Generating Ad Angles

• Triggered by a web hook when "Generate Angles" is selected in Air Table.
• Retrieves product information, including product name, details, notes, and winning script (if available).
• Constructs a prompt for Google Gemini 2.5 Pro, instructing it to generate multiple ad angles with specific details (concepts, VSSL structures, video suggestions, emotional triggers, cognitive biases, direct response techniques).
• Uses a parsing module to structure the output into an array.
• Relates the generated ad angles to the specific product in Air Table.
• Updates the Air Table record with the generated angles.

3.3. Generating Scripts and Handling Feedback

• Triggered by a web hook when script generation is initiated.
• Retrieves angle details and product information.
• Uses Google Gemini 2.5 Pro to generate the VSSL script based on the provided information.
• Updates the Air Table record with the generated script.
• Handles script revisions by incorporating user feedback and regenerating the script.

3.4. Scraping Best Performing Ads

• Triggered by a web hook when a "best ad" is selected in Air Table.
• Retrieves the video ID from Air Table.
• Uses the Facebook Graph API to retrieve the video's perma link.
• Scrapes the Facebook page using Scrapfly to find the direct video URL.
• Extracts the video URL using a code step.
• Uses an HTTP request to scrape the video content.
• Transcribes the video using an OpenAI transcribing module.
• Outputs the transcript back to Air Table.

3.5. Ad Performance Analytics

• Uses a scheduled trigger to run periodically (e.g., weekly).
• Uses the Facebook Graph API to retrieve ad account, campaign, ad set, and ad content data.
• Filters for active ad accounts, campaigns, and ad sets.
• Filters for video ads.
• Maps relevant data points (ad spend, CPC, clicks, conversions, etc.) to Air Table.

4. Facebook Graph API Integration

• Requires creating an app on developers.facebook.com.
• Selecting "Other" as the use case and "Business" as the app type.
• Choosing the appropriate business portfolio or ad account.
• Setting up the Marketing API and obtaining an access token.
• Using the access token to connect to the Facebook Graph API in NAN.

5. Web Scraping with Scrapfly

• Scrapfly is used to scrape the Facebook page and extract the direct video URL.
• Scrapfly is considered a robust web scraper for handling complex websites.

6. AI Models

• Perplexity AI (Sonar Reasoning model): Used for deep research on product details and customer profiles.
• Google Gemini 2.5 Pro: Used for creative writing tasks, such as generating ad angles and scripts.

7. Key Arguments and Benefits

• Automation Saves Time and Resources: Automating the ad creation process significantly reduces the workload for ad agencies.
• Improved ROAS: By leveraging data from top-performing ads, the system helps optimize ad campaigns and increase return on ad spend.
• Scalability: The system allows agencies to manage multiple clients and scale their ad operations efficiently.
• Data-Driven Optimization: The system uses data from ad performance to continuously improve and optimize ad campaigns.

8. Conclusion

The video presents a comprehensive system for automating Meta ad agency operations using Air Table and NAN. By integrating AI-powered research, creative generation, and performance tracking, the system streamlines the ad creation process, improves ROAS, and enables scalability. The template is available for free download, offering a valuable resource for ad agencies looking to automate their workflows.