Technical Guide

How to Format Product Data for AI Systems

AI systems need data in specific formats to process it correctly. This technical guide covers file formats, data structures, and formatting best practices.

Quick Answer

CSV format is best for beginners - simple, spreadsheet-compatible, easy to maintain. Use UTF-8 encoding, store dimensions in centimeters (cm) as numeric values, format multi-value fields with pipe separators (color1|color2), and provide high-resolution JPG/PNG images (min 1000px) with direct URLs. Keep data clean: no HTML, no embedded units, consistent headers.

CSV for simple data Dimensions in cm UTF-8 encoding High-res images

Accepted File Formats

Fringe accepts product data in several formats. Choose based on your technical capabilities:

๐Ÿ“Š

CSV/Excel

Best for: Non-technical teams. Simple spreadsheet format.

Recommended for beginners

๐Ÿ“‹

JSON

Best for: Technical teams. Supports nested data structures.

For complex product data

๐Ÿ”—

API Integration

Best for: Real-time sync. Automated updates from your PIM.

For enterprise clients

CSV Format Guidelines

CSV (Comma-Separated Values) is the simplest format. Here's how to structure it:

# Example CSV structure

sku,name,brand,category,width_cm,depth_cm,height_cm,material,price,currency,image_url
MIL-001,"Milano Chair","DesignCo","Seating > Lounge",80,75,82,"leather",1500,"EUR","https://..."
MIL-002,"Milano Sofa","DesignCo","Seating > Sofas",200,90,82,"leather",3500,"EUR","https://..."

CSV Best Practices

  • โ€ข Use UTF-8 encoding for international characters
  • โ€ข Quote text fields that may contain commas
  • โ€ข Use consistent column headers across updates
  • โ€ข Include header row as first line
  • โ€ข Use blank cells for missing data (not "N/A" or "-")
  • โ€ข Save as .csv, not .xlsx

JSON Format Guidelines

JSON allows for more complex data structures with nested objects:

{
  "products": [
    {
      "sku": "MIL-001",
      "name": "Milano Chair",
      "brand": "DesignCo",
      "category": ["Seating", "Lounge Chairs"],
      "dimensions": {
        "width_cm": 80,
        "depth_cm": 75,
        "height_cm": 82,
        "seat_height_cm": 45
      },
      "materials": {
        "primary": "full-grain leather",
        "frame": "solid oak"
      },
      "colors": ["black", "cognac", "navy"],
      "price": {
        "amount": 1500,
        "currency": "EUR"
      },
      "certifications": ["GREENGUARD", "FSC"],
      "applications": ["hospitality", "residential", "office"],
      "images": [
        {"url": "https://...", "type": "primary"},
        {"url": "https://...", "type": "detail"}
      ]
    }
  ]
}

Data Type Guidelines

Data Type Format Example
Text Plain text, quoted if commas "Milano Chair"
Numbers No quotes, use decimals 80.5
Boolean true/false or 1/0 true
Lists (CSV) Pipe-separated in single field "black|cognac|navy"
Lists (JSON) Array format ["black", "cognac"]
URLs Full URL with https:// "https://example.com/img.jpg"
Prices Number only, currency separate 1500 (not "โ‚ฌ1,500")

Dimension Standards

Use consistent units across all products:

  • Length/Width/Height: Centimeters (cm) - not inches or mm
  • Weight: Kilograms (kg)
  • Capacity: Kilograms (kg) for weight, liters (L) for volume
  • Area: Square meters (mยฒ) for larger surfaces

Image Requirements

Do

  • โœ“ Min 1000px on shortest side
  • โœ“ White or neutral background
  • โœ“ JPG or PNG format
  • โœ“ Direct image URLs (not page URLs)
  • โœ“ Multiple angles if available

Don't

  • โœ— Watermarked images
  • โœ— Lifestyle images as primary
  • โœ— Images with text overlays
  • โœ— Low resolution (<500px)
  • โœ— Expired/broken URLs

Common Formatting Mistakes

Inconsistent units

Mixing cm and inches, or including units in numeric fields ("80 cm" instead of 80)

Embedded formatting

Price as "โ‚ฌ1,500.00" instead of just "1500" with separate currency field

HTML in text fields

Including <br> tags or HTML formatting in descriptions

Encoding issues

Special characters appearing as "รƒยค" instead of "รค" - use UTF-8

Need a Data Template?

Contact us to receive a pre-formatted template for your product category.

Get Template

Frequently Asked Questions

Quick answers to common questions about this topic

Frequently Asked Questions

What file format is best for AI product data?

CSV (Comma-Separated Values) is best for beginners and non-technical teams due to its simplicity and spreadsheet compatibility. JSON is recommended for technical teams who need nested data structures and complex product relationships. For enterprise clients with real-time requirements, API integration is the optimal choice. All formats work with AI systems, so choose based on your team's technical capabilities and data complexity.

Should I use CSV, JSON, or XML for product data?

CSV is ideal for simple product catalogs with flat data structures and is easy to edit in Excel or Google Sheets. JSON is better for complex products with nested attributes like multiple materials, dimensions, or variant options. XML is less common but supported if your existing systems use it. Most teams start with CSV and upgrade to JSON as their data complexity grows.

How should I format dimensions for AI?

Use centimeters (cm) for length, width, and height measurements. Use kilograms (kg) for weight. Always use numeric values only without units in the data field - store 80 not '80 cm'. Keep units consistent across all products. Create separate fields for each dimension: width_cm, depth_cm, height_cm rather than combining them into a single field.

What image formats does AI prefer?

AI systems work best with JPG or PNG images that are at least 1000px on the shortest side. Use clean product shots with white or neutral backgrounds. Provide direct image URLs (ending in .jpg or .png) rather than webpage URLs. Include multiple angles when available. Avoid watermarked images, text overlays, or low-resolution photos under 500px.

How do I format multi-value fields like colors?

In CSV files, use pipe separators to list multiple values in a single field: 'black|cognac|navy'. In JSON files, use array format: ['black', 'cognac', 'navy']. Never use commas in CSV multi-value fields as this breaks the CSV structure. Maintain consistent terminology - don't mix 'brown' and 'cognac' for the same color across products.

What are the most common data formatting mistakes?

The most common mistakes are: mixing measurement units (cm and inches), including units in numeric fields ('80 cm' instead of 80), embedding formatting in prices ('โ‚ฌ1,500' instead of 1500), using HTML tags in text fields, character encoding issues from not using UTF-8, and inconsistent column headers between data updates. Always validate your data before submission.

Ready to Make Your Products AI-Discoverable?

Join thousands of interior design and architecture suppliers who are already reaching designers through AI-powered search.

Continue Reading