Replicate Competitor Ads with n8n – Technical & Ethical Implementation Guide

This guide documents a production-ready n8n workflow template (The Recap AI – Facebook Ad Thief) that automates the process of analyzing Facebook ads and generating new ad images that feature your own product. It explains the workflow architecture, node-by-node behavior, configuration requirements, and ethical constraints so that advanced users can deploy and adapt it safely.

1. Workflow Overview

The n8n template automates a full pipeline from Facebook Ad Library URL input to generated ad creatives stored in Google Drive. At a high level, the workflow:

• Receives a Facebook Ad Library URL and a product image via a form trigger.
• Invokes an Apify actor to scrape ad metadata and images from the provided URL.
• Downloads the reference ad images, archives them to Google Drive, and converts them to base64.
• Combines the reference images with your product image to build a structured prompt.
• Calls Google Gemini generative endpoints through HTTP Request nodes to:
  • Generate or refine an image prompt.
  • Create new ad images that visually resemble the reference ad but feature your product and branding.
• Performs prohibited-content checks on the model output.
• Converts the generated base64 images back to binary and uploads them to Google Drive.

The objective is to accelerate creative experimentation and A/B testing by programmatically generating ad variants that are inspired by competitor ads, while intentionally replacing competitor branding with your own.

2. System Architecture & Data Flow

The workflow orchestrates multiple external services and n8n nodes. The data flow is roughly:

1. Input collection
   • User submits:
     • Facebook Ad Library URL (source ad).
     • Product image (your brand asset).
2. Ad scraping
   • n8n calls an Apify Facebook Ad Library scraper actor with the URL.
   • Actor returns ad metadata and image URLs.
3. Reference image handling
   • Workflow downloads the ad images.
   • Uploads them to a designated Google Drive folder for archival.
   • Converts images to base64 for use in Gemini API requests.
4. Prompt construction
   • Product image is also converted to base64.
   • Reference and product images are aggregated.
   • A structured text prompt is generated to instruct Gemini how to:
     • Replace competitor branding.
     • Preserve layout and style.
     • Maintain ad copy and calls to action, except for brand-specific elements.
5. Image generation
   • HTTP Request nodes call Google Gemini (gemini-2.5 endpoints) to:
     • Draft or refine the image-generation prompt.
     • Generate one or more ad image variants.
6. Safety checks
   • Response is inspected for prohibited content flags.
   • Flagged outputs are not stored or used.
7. Output storage
   • Valid generated images are decoded from base64.
   • Binary files are created and uploaded to an output folder in Google Drive.

3. Node-by-Node Breakdown

3.1 Entry & Input Handling

form_trigger

• Role: Primary trigger for the workflow.
• Inputs collected:
  • Facebook Ad Library URL (string field).
  • Product image (file upload field).
• Behavior: When a user submits the form, n8n starts the workflow and passes both the URL and the file reference to downstream nodes.

convert_product_image_to_base64

• Role: Convert the uploaded product image into a base64-encoded string.
• Reason: Image-generation APIs typically accept image data as base64 within JSON payloads rather than as multipart file uploads.
• Output: A base64 string representing your product image, used later in prompt construction and Gemini requests.

3.2 Scraping & Reference Image Preparation

scrape_ads (Apify actor)

• Type: Apify integration node calling a Facebook Ad Library scraper actor.
• Credentials: Requires configured Apify OAuth credentials.
• Key parameters:
  • Actor ID: Set to the Facebook Ad Library scraper actor ID.
  • Input: Includes the Facebook Ad Library URL from form_trigger.
  • limitPerSource: Controls the number of ads scraped per URL.
• Output: JSON structure containing:
  • Ad metadata (e.g., creative details, cards).
  • Image URLs for each scraped ad.
• Notes:
  • If the URL is invalid or the actor fails, this node will error. Handle by testing with a single known-good URL first.
  • Respect Apify usage limits and Facebook terms of service.

iterate_ads

• Role: Iterate through each ad returned by the Apify actor.
• Behavior:
  • Loops over the list of scraped ads.
  • For each ad, extracts the primary image URL or card image URL.
• Output: One execution per ad, with ad-level context passed to downstream nodes.

download_image

• Type: HTTP Request or dedicated file-download node, depending on template version.
• Role: Download the reference image from the ad image URL.
• Output: Binary image data for the reference ad image.
• Edge cases:
  • Broken or expired URLs will cause download failures.
  • Non-image MIME types should be treated as errors or filtered out.

upload_ad_reference (Google Drive)

• Role: Archive the downloaded reference ad image to Google Drive.
• Credentials: Uses Google Drive OAuth credentials.
• Configuration:
  • Target folder: A dedicated folder for reference ads.
  • File name: Often derived from ad ID or timestamp for traceability.
• Purpose: Maintain a record of the original reference creatives for auditing and later review.

convert_ad_image_to_base64

• Role: Convert each downloaded reference ad image into base64.
• Output: Base64-encoded string for each reference ad image, used as part of the Gemini prompt context.

3.3 Aggregation & Prompt Construction

aggregate

• Role: Aggregate multiple inputs:
  • Product image (base64) from convert_product_image_to_base64.
  • Reference ad images (base64) from convert_ad_image_to_base64.
• Behavior:
  • Combines data into a single structure that can be used to construct a unified prompt.
  • May limit the number of reference images to avoid very large payloads.

build_prompt

• Role: Generate the textual instructions that will be sent to the image-generation model.
• Prompt strategy:
  • Brand replacement rules:
    • Explicitly instruct the model to replace competitor brand text (for example, “AG1”) with your brand (for example, “ThriveMix”).
    • Include partial-token handling such as “A”, “AG”, or “G1” so that partial branding is also removed or replaced.
  • Style and layout preservation:
    • Describe lighting, shadows, color grading, and overall mood to match the original ad.
    • Instruct the model to maintain composition and layout where possible.
  • Product placement rules:
    • If multiple products appear in the reference image, instruct the model to place your product in the primary, visually dominant position.
  • Copy and CTA handling:
    • Retain the original ad copy and call-to-action structure.
    • Replace only the brand names and brand-specific terms with your own.
• Output: A finalized text prompt string, plus references to the base64 images that will be provided to Gemini.

3.4 Gemini Integration (Prompt & Image Generation)

generate_ad_image_prompt (Google Gemini)

• Type: HTTP Request node.
• Role: Call Gemini generative language or multimodal endpoint to refine or validate the prompt before image generation.
• Endpoint: Configured to use gemini-2.5 API endpoints.
• Authentication:
  • Uses a Google Cloud API key or similar credential configured in n8n via httpHeaderAuth.
• Behavior:
  • Sends the constructed prompt and relevant context.
  • Receives a refined prompt or additional structured instructions to improve final image quality.

generate_ad_image (Google Gemini)

• Type: HTTP Request node.
• Role: Request final image generation from Gemini using the refined prompt and base64 images.
• Endpoint: Gemini image-generation endpoint for the gemini-2.5 model family.
• Inputs:
  • Text prompt from build_prompt or generate_ad_image_prompt.
  • Base64-encoded product and reference images.
• Output:
  • Base64-encoded generated image(s).
  • Metadata that may include safety or content flags.
• Edge cases:
  • Large base64 payloads can cause request size issues.
  • Unsupported image dimensions or formats may result in API errors.

3.5 Safety & Output Storage

check_if_prohibited

• Role: Evaluate Gemini’s response for prohibited or sensitive content.
• Inputs:
  • Model safety metadata or flags from the Gemini response.
• Behavior:
  • If the result indicates disallowed content, the workflow branches to a path that skips storage and downstream use.
  • Otherwise, execution continues to image decoding and upload.
• Purpose: Enforce platform and internal safety policies before any asset is persisted or used in campaigns.

get_image

• Role: Convert base64-encoded generated images back into binary format.
• Output: Binary image files suitable for upload to Google Drive or other storage systems.

upload_image (Google Drive)

• Role: Store generated ad images in a dedicated Google Drive folder.
• Credentials: Uses the same Google Drive OAuth credentials as the reference upload node.
• Configuration:
  • Output folder: A separate folder for generated creatives, distinct from reference ads.
  • File naming: Can include timestamp, campaign ID, or source ad identifier.
• Result: A collection of generated ad images ready for manual review and potential deployment.

4. Setup & Configuration Checklist

4.1 Prerequisites

• Running n8n instance (self-hosted or cloud).
• Access to:
  • Apify account with Facebook Ad Library scraper actor.
  • Google Drive API with OAuth credentials.
  • Google Cloud project with access to Gemini generative endpoints.

4.2 Installation Steps

1. Import the template
   • Install n8n.
   • Import the provided workflow JSON into your n8n canvas.
2. Configure Apify
   • Create or configure Apify OAuth credentials in n8n.
   • Open the scrape_ads node and set:
     • Actor ID to the Facebook Ad Library scraper actor.
     • Any required input fields for the actor (including the URL from form_trigger).
   • Set limitPerSource to control how many ads you scrape per run.
3. Configure Google Drive
   • Create Google Drive OAuth credentials in n8n.
   • In upload_ad_reference, set:
     • Credential to the new Google Drive OAuth.
     • Folder ID or path for storing reference ads.
   • In upload_image, set:
     • Same or different credential, as needed.
     • Folder ID or path for storing generated creatives.
4. Configure Gemini API access
   • In Google Cloud, provision an API key or appropriate credential that can call Gemini gemini-2.5 endpoints.
   • In n8n, create an HTTP Header Auth or