Product Data Requirements

For optimal performance of Youzu.ai's visual AI features, it's important to ensure your product data meets our recommended specifications. This guide outlines the requirements and best practices for structuring your product data.

Required Fields

All product records must include these essential fields:

Field Type Description Example
product_id String Unique identifier for the product "prod_12345"
name String Product name "Ergonomic Office Chair"
description String Product description "Premium office chair with lumbar support"
price Number Product price 299.99
currency String Three-letter currency code "USD"
images Array At least one product image See Image Requirements

Image Requirements

Images are crucial for Youzu.ai's visual AI functionality. For best results:

Technical Specifications

  • Resolution: Minimum 800x800 pixels, recommended 1200x1200 pixels or higher
  • Format: JPEG, PNG, or WebP
  • File size: Maximum 10MB per image
  • Background: White or transparent backgrounds preferred
  • Quantity: At least one image per product, multiple images recommended

Image Quality Best Practices

  • Use well-lit, professional product photography
  • Show the product from multiple angles (front, side, back, detail shots)
  • Avoid text overlays or watermarks on the image
  • Ensure the product fills at least 80% of the image frame
  • Maintain consistent lighting and background across product images

Image Object Structure

When using JSON format, images should be structured as an array of objects:

"images": [
  {
    "url": "<https://example.com/images/chair_front.jpg>",
    "position": 1,
    "alt": "Front view of office chair"
  },
  {
    "url": "<https://example.com/images/chair_side.jpg>",
    "position": 2,
    "alt": "Side view of office chair"
  }
]

When using CSV format, you can include a primary image in the image_url column and additional images in numbered columns:

image_url,image_url_2,image_url_3
<https://example.com/images/chair_front.jpg>,<https://example.com/images/chair_side.jpg>,<https://example.com/images/chair_back.jpg>

While not required, these fields significantly improve the performance of Youzu.ai's visual search and recommendation features:

Field Type Description Example
categories Array/String Product categories ["Furniture", "Office"]
tags Array/String Product tags ["ergonomic", "office", "comfortable"]
variants Array Product variants See Variant Structure
brand String Product brand "ErgoDesigns"
metadata Object Additional product details {"weight": "15kg", "dimensions": "60x70x120cm"}

Variant Structure

If your products have variants (e.g., different colors, sizes), structure them as follows:

"variants": [
  {
    "id": "var_789",
    "attributes": {
      "color": "Black",
      "material": "Leather",
      "size": "Standard"
    },
    "price": 299.99,
    "image_url": "<https://example.com/images/chair_black.jpg>"
  },
  {
    "id": "var_790",
    "attributes": {
      "color": "Brown",
      "material": "Fabric",
      "size": "Standard"
    },
    "price": 279.99,
    "image_url": "<https://example.com/images/chair_brown.jpg>"
  }
]

Character Limits

To ensure optimal processing, adhere to these character limits:

Field Maximum Length Notes
product_id 50 characters Alphanumeric characters, hyphens, and underscores
name 200 characters Brief, descriptive title
description 5,000 characters Detailed product information
categories 50 characters per category Keep categories concise and consistent
tags 30 characters per tag Use relevant, searchable terms

Data Validation

Before submitting your product data to Youzu.ai, verify:

  1. All required fields are present
  2. Image URLs are valid and accessible
  3. Product IDs are unique within your catalog
  4. Prices are formatted as numbers without currency symbols
  5. JSON is properly formatted with no syntax errors

Field Name Flexibility with Property Mapping

Youzu.ai recognizes that different e-commerce platforms and internal systems may use different naming conventions for product attributes. To accommodate this, we provide a property mapping feature that allows you to map your existing property names to our expected field names.

For example, if your system uses item_id instead of product_id, you can create a property map to automatically translate these fields during import without changing your source data.

Your System Youzu.ai Expected
item_id product_id
title name
long_description description
retail_price price