The 35,000-Product Catalog That Publishes Itself

The situation

The client wanted a dropshipping store with more than 35,000 products from one large supplier. The supplier's API did not include everything a product page needs, such as full descriptions, weights, and categories, so the missing details came from a very large CSV file. Entering that many products by hand, writing Danish descriptions, and keeping a clean category tree was out of the question.

What we built

A product pipeline in connected steps. A Python script reads the 35,000+ product CSV and loads the data into Supabase, which acts as one central product database. An n8n workflow then works through the products on a schedule. It uses AI to translate and improve each description into Danish and to help place the product in a category, while keeping the original product information.

The workflow also manages the store's categories. Before it creates a new parent or child category in WooCommerce, it checks whether one already exists, so the catalog stays clean. Finished products are converted into the format WooCommerce expects and published with descriptions, images, prices, categories, and supplier tags.

How it works

SEE IT IN MOTIONPause

CSVSupplier CSVProduct details

PythonReads the file

SupabaseProduct database

n8nRuns on a schedule

OpenAITranslates, sorts

WooCommerceYour store

1234

Step 1 of 4A script reads the large supplier CSV and organizes it in a central product database.

An illustration of the steps below. Example flow only, no client data.

1. 01A script reads the large supplier CSV and organizes it in a central product database.
2. 02A scheduled n8n workflow picks up products from the database.
3. 03AI translates and tidies each description into Danish and helps place the product in a category. The workflow checks the store's existing categories before creating any.
4. 04The finished product is published to WooCommerce with its images, price, categories, and supplier tag.

The challenges we solved

• The supplier's API did not hold all the product data, so the missing details had to come from a huge CSV before anything could be published.
• Thousands of products meant thousands of category decisions, so the workflow had to reuse existing parent and child categories instead of creating duplicates.
• Descriptions had to read naturally in Danish while keeping the supplier's original product facts intact.

The result

The store's catalog is now filled by a pipeline instead of a person. More than 35,000 products can be processed and published with Danish descriptions, clean categories, and the right images and prices, and the workflow keeps running on a schedule.