Product Enrichment

Automatically enhance your product catalog with detailed attributes extracted by AI from web sources.

Overview

Product Enrichment helps you build comprehensive product catalogs by automatically extracting and organizing product information. The AI analyzes web pages and generates structured attributes for your products, saving you hours of manual data entry.

Getting Started

Adding Products to Your Catalog

Before enriching, you need products in your catalog. You have three options:

Option 1: Add Single Product

1. Navigate to Products from the main menu
2. Click Add Product button
3. Fill in basic information:
   • Product Name (required)
   • SKU (required)
   • Brand, Manufacturer, or MPN (optional but helpful)
4. Click Save Product

Option 2: Import from CSV

1. Navigate to Products from the main menu
2. Click Import Products
3. Upload a CSV file with your product data
4. Supported columns: sku, name, manufacturer, mpn, uom, brand
5. Click Import - duplicates will be skipped automatically

Option 3: Use the API

Programmatically add products via our REST API for automated workflows.

Updating Existing Products from CSV

Use Update Products via CSV when you need to change many existing products without editing them one at a time. You can update core product fields, parent-child relationships, data sources, and configured enrichment attributes in the same file.

Upload an Update File

1. Go to Products
2. Click Bulk Actions → Update Products via CSV
3. Select or drop a .csv file
4. Click Update Products

Each row needs an identifier for an existing product. The recommended identifiers are:

When a row contains more than one identifier, the internal ID takes priority, followed by external ID and then SKU. Use an internal ID when you need to change the product's external ID. Use an internal ID or external ID when you need to change its SKU.

Updateable Columns

Only columns included in the uploaded file are updated. Omitted columns remain unchanged.

Match by External ID and Change the SKU

external_id,sku,name,brand
ERP-10001,SKU-10001-V2,Updated Wireless Headphones,AudioCorp
ERP-10002,SKU-10002-V2,Updated Wireless Mouse,InputWorks

In this example, external_id identifies each existing product. The SKU, name, and brand are updated; the external ID remains unchanged.

Match by Internal ID and Change the External ID

id,external_id,sku,name
product_abc123,ERP-10001-V2,SKU-10001-V2,Updated Wireless Headphones
product_def456,ERP-10002-V2,SKU-10002-V2,Updated Wireless Mouse

You can use IDs from a product export as the id values in an update file.

Skipped Rows

A row is skipped when, for example:

• No matching product exists in the active organization
• The row has no usable identifier or updateable columns
• The new SKU or external ID belongs to another product
• The same product appears more than once in the file
• is_parent is invalid, the parent product cannot be found, or the relationship would make a product its own parent
• A product would be both a parent and a child

When the file contains at least one valid row, valid rows are still updated even if other rows are skipped. After a successful batch with skipped rows, the browser downloads a CSV containing the original row data and a Skip Reason column. Fix those rows and upload the downloaded file again.

Enriching Products

Bulk Enrichment Process

Step 1: Select Products

• Check the boxes next to products you want to enrich
• Or use Select All to choose all products on the page
• You can select products across multiple pages
• Use the Type filter to select only Parent Products, Non-Parent Products, or All Products

Step 2: Start Enrichment

• Click Enrich Selected button in the toolbar
• A model selection dialog will appear

Step 3: Choose AI Model

Select the model that best fits your needs:

• Budget - Fast and economical for basic attributes
• Normal - Recommended balance of quality and cost
• Pro - Enhanced accuracy for detailed specifications
• Ultra - Premium quality for complex products

Step 4: Configure Options

• Extract Product Images: Toggle to automatically extract and store product images from web sources during enrichment
  • When enabled, the system will search for product images and store them for later use
  • Extracted images are categorized by priority: primary, secondary, or detail images
  • Images are saved alongside enrichment data for easy access
• Include existing attributes: On by default. Current attribute values are sent to the model as a starting point
  • Turn this off when re-enriching a product so the model regenerates from sources instead of keeping old specs
• Use specific attribute set: Check this option to use a custom Attribute Set instead of all attributes
  • Select an attribute set from the dropdown
  • Only attributes in the selected set will be used for enrichment
  • Prompt overrides defined in the attribute set will be applied

Step 5: Confirm

• Review the estimated credit cost (image extraction may incur additional credits)
• Click Confirm & Enrich to start the job

Step 6: Monitor Progress

• A real-time progress notification appears in the corner
• Track job status on the Products page
• View detailed progress in Job History → Enrichment Jobs
• See extraction settings for each job in the job history table

Single Product Enrichment

You can also enrich individual products:

From Products List:

1. Click the ⋮ (three dots) menu on any product row
2. Select Enrich Product
3. Choose your AI model
4. Click Confirm & Enrich

From Product Detail Page:

1. Click on any product to open its details
2. Click Enrich Product button in the header
3. Select your model and confirm

Parent Product Enrichment

Parent products (products marked as having variants) are enriched differently using intelligent data aggregation from their child products:

How It Works:

• The system analyzes all child product variants
• Aggregates common attributes shared across children
• Creates ranges for numeric values (e.g., "10-20 kg" for weight)
• Combines categorical values (e.g., "Red, Blue, Green" for colors)
• Summarizes descriptive attributes based on all variants

Aggregation Strategies:

1. Common/Shared Values

   • If all children have the same value → Parent uses that value
   • Examples: Brand, Manufacturer, Category

2. Range Values (Numeric)

   • If children have different numbers → Parent shows range
   • Format: "[minimum] - [maximum] [unit]"
   • Examples: Price ranges, Weight ranges, Dimension ranges

3. Combined Values (Categorical)

   • If children have different options → Parent lists all unique values
   • Examples: Available Colors, Available Sizes

4. Summarized Values (Descriptive)

   • For text descriptions → Parent gets a comprehensive summary
   • Describes common features and mentions variations

Requirements:

• Parent product must have at least one child product linked to it
• Child products should have enriched attributes for best results
• During enrichment, the system will analyze all child variants automatically

Benefits:

• Automatically maintain parent product data based on variants
• Ensure parent attributes accurately represent the product family
• Save time by not manually updating parent products
• Keep parent data synchronized with child variations

Example:

If you have a "Classic T-Shirt" parent product with child variants:

• Red T-Shirt Size S (Price: $19.99, Weight: 200g)
• Blue T-Shirt Size M (Price: $22.99, Weight: 220g)
• Green T-Shirt Size L (Price: $24.99, Weight: 240g)

The parent product enrichment will automatically create:

• Colors: "Red, Blue, Green"
• Sizes: "S, M, L"
• Price: "$19.99 - $24.99"
• Weight: "200g - 240g"
• Brand: (same value if all children share it)

Note: Parent product enrichment does not use web search. It relies entirely on aggregating data from child products, so it consumes fewer credits than standard enrichment.

AI Model Comparison

Budget Model

• Best For: Large catalogs, standard products
• Processing Time: ~5-10 seconds per product
• Credit Cost: Lowest (approximately 0.05-0.10 credits)
• Accuracy: Good for basic information
• Example Use: Consumer goods, simple products, bulk enrichment

Normal Model (Recommended)

• Best For: Most products and general use cases
• Processing Time: ~15-20 seconds per product
• Credit Cost: Moderate (approximately 0.15-0.30 credits)
• Accuracy: High for standard attributes
• Example Use: E-commerce catalogs, retail products

Pro Model

• Best For: Technical products, detailed specifications
• Processing Time: ~30-45 seconds per product
• Credit Cost: Higher (approximately 0.40-0.60 credits)
• Accuracy: Very high with complex attribute extraction
• Example Use: Electronics, industrial equipment, machinery

Ultra Model

• Best For: Mission-critical accuracy requirements
• Processing Time: ~60-90 seconds per product
• Credit Cost: Highest (approximately 0.80-1.20 credits)
• Accuracy: Best-in-class extraction and validation
• Example Use: Medical devices, specialized equipment, regulated products

Note: Credit costs vary based on product complexity and attribute count

Viewing Enrichment Results

From Products List

1. Look for the eye icon (👁️) next to enriched products
2. Click View Attributes from the dropdown menu
3. Browse all extracted attributes in a dialog

From Product Detail Page

1. Click on any product to view its details
2. All enriched attributes appear in organized sections
3. Edit attributes directly if needed

Enrichment History

Each product maintains a complete history:

• Navigate to any product detail page
• Scroll to Enrichment History section
• View all past enrichment jobs with:
  • Enriched At: When the enrichment was completed
  • Model Type: Which AI model was used (Budget, Normal, Pro, Ultra)
  • Extract Images: Whether image extraction was enabled (Yes/No)
  • Attributes: Number of attributes extracted
• Review extracted attributes for each run
• Click the eye icon to view detailed enrichment results

Job Management

Viewing Job History

Access your enrichment job history:

1. Go to Products page
2. Click Job History → Enrichment Jobs
3. View all past and current jobs

Job Information

For each job, you can see:

• Job ID: Unique identifier for tracking
• Status: Current state (Pending, Processing, Completed, etc.)
• Extract Images: Whether image extraction was enabled for this job (Yes/No)
• Progress: X out of Y products completed with progress bar
• Success Count: Number of products enriched successfully
• Failed Count: Number of products that encountered errors
• Created By: User who started the job
• Timestamps: Creation, start, and completion times

Job Statuses

Viewing Job Details

Click the View button on any job to see:

• Complete list of all products in the job
• Individual status for each product
• Extracted attributes for successful products
• Error messages for failed products
• Pagination for large jobs (50 items per page)
• Extract Images: Shows whether image extraction was enabled for this job and the AI model used

Product Image Extraction

Overview

Product Image Extraction is an optional feature that automatically discovers and stores product images from web sources during the enrichment process. When enabled, the system searches the web for product images and categorizes them by priority.

Enabling Image Extraction

1. Start an enrichment job as usual
2. In the model selection dialog, enable the Extract Product Images toggle
3. The toggle appears only for enrichment jobs (not validation)
4. Images will be automatically extracted and stored alongside your product data

How Extracted Images Work

When image extraction is enabled:

• The system searches for product images from available web sources
• Images are categorized by priority:
  • Primary: Main product images (usually the official product photo)
  • Secondary: Alternative product views (different angles, colors)
  • Detail: Close-up or detailed shots (specifications, features)
• Each image includes:
  • Image URL
  • Alt text (description)
  • Priority classification
  • Source website information

Image Storage

Extracted images are stored:

• Automatically: Alongside enriched product attributes
• Persistently: Available for future reference and downloads
• Organized: By website source for easy identification
• Searchable: Through the enrichment history

Viewing Extracted Images

1. After enrichment completes, go to the product detail page
2. Scroll to the Enrichment History section
3. Click on a completed enrichment job
4. View extracted images with their priority and source information

Viewing All Extracted Images

On the product detail page, you can view all extracted images in one place:

1. Open any product detail page
2. Click the Images tab in the navigation menu
3. View all extracted images organized by source website
4. Each image displays:
   • Priority Badge: Visual indicator of image type (Primary, Secondary, Detail)
   • Website Source: Which website the image came from
   • Alt Text: Image description
   • Clickable Link: Click to view the original image in full size

Priority Labels Explained

When viewing extracted images, priority badges help you understand the image type:

• Primary: Main product images (official photos, best quality)
• Secondary: Alternative product views (different angles, colors, variations)
• Detail: Close-up shots (specifications, features, detailed views)

Troubleshooting Image Viewing

Image Extraction Best Practices

• Test First: Run image extraction on a small batch to verify results
• Combined with Models: Use with Pro or Ultra models for better accuracy
• Large Catalogs: Consider enabling for high-value products where visual representation matters
• Bulk Operations: Image extraction can increase processing time - factor this into schedules

Troubleshooting Image Extraction

Product Attributes

What Are Attributes?

Attributes are structured pieces of information about your products. Examples include:

• Product descriptions
• Dimensions and weight
• Materials and colors
• Technical specifications
• Features and benefits
• Compatibility information
• Warranty details

Viewing Attributes

All enriched attributes appear in a clean, organized format:

• Text attributes display as formatted paragraphs
• Lists appear as bullet points
• Numbers show with proper units
• Boolean values display as Yes/No

Editing Attributes

After enrichment, you can manually refine any attribute:

1. Open the product detail page
2. Click the Edit icon next to any attribute
3. Make your changes
4. Save the updated information

Creating Custom Attributes

1. Go to Settings → Enrichment Settings
2. Click Add Attribute
3. Fill in the details:
   • Name: Human-readable attribute name (e.g., "Product Description")
   • Slug: URL-friendly identifier (e.g., "product_description")
   • Type: Data type - text, number, boolean, array
   • Prompt: AI instructions describing what to extract (default for all products)
   • Category-Specific Prompts: Optional override prompts for specific product categories
   • Options: Predefined values for structured attributes (optional)
   • Order: Attribute sequence

Setting Category-Specific Prompts:

Click Add Category Prompt to define extraction instructions for specific categories:

1. Enter the category name (e.g., "Electronics > Computers")
2. Enter the category-specific prompt
3. Click Add to save

You can add multiple category-specific prompts for the same attribute. During enrichment, if a product's category matches one of your defined categories, that prompt will be used instead of the default.

Best Practices for Prompts

Good Prompts:

"Extract the product dimensions in format: W x H x D (include units)"
"List all available colors as a comma-separated list"

Poor Prompts:

"Get size" - Too vague
"Dimensions" - No output format specified

Category-Specific Prompt Tips:

1. Be Specific: Tailor prompts to the unique characteristics of each category
2. Use Examples: Include format examples relevant to that category
3. Consider Context: Think about what information matters most for that product type
4. Keep Consistency: Maintain similar output formats across categories when possible

Example: "Description" Attribute

• Default: "Write a comprehensive product description highlighting key features and benefits"
• Electronics > Smartphones: "Write a description emphasizing display technology, camera capabilities, processor, battery life, and operating system"
• Food & Beverages: "Write a description focusing on ingredients, nutritional benefits, flavor profile, and usage suggestions"
• Furniture: "Write a description highlighting materials, dimensions, assembly requirements, style, and care instructions"

Product Categories

Products can be automatically categorized using a standardized product taxonomy (6,000+ categories). Categories help organize products and improve the quality of AI enrichment through category-specific attribute extraction.

Benefits of Categorization

• Consistent Organization: Standard categories across your entire catalog
• Better Discovery: Easier to filter and find products
• E-commerce Ready: Categories compatible with major e-commerce platforms and product feeds
• Improved SEO: Better search engine understanding of your products
• Category-Specific Enrichment: AI uses category-specific prompts for more accurate attribute extraction

How It Works

There are two ways to set product categories:

1. Automatic Categorization (During Enrichment)

The AI suggests appropriate categories during enrichment based on product characteristics. You can accept or change these suggestions on the product detail page.

2. Manual Categorization (Before Enrichment)

You can set categories manually before enriching products:

• When creating products via API, include the category field
• When importing products via CSV, include a category column
• Edit the category on the product detail page

Category Format:

Categories follow a hierarchical taxonomy format:

Electronics > Audio > Headphones
Home & Garden > Kitchen > Appliances > Coffee Makers
Apparel & Accessories > Clothing > Activewear

Category-Specific Enrichment Prompts

When configuring enrichment attributes in Settings → Enrichment Settings, you can define category-specific prompts. This allows different instructions for extracting the same attribute based on the product's category.

Example Use Case:

For a "Key Features" attribute:

• Default Prompt: "List the main features of this product"
• Electronics > Computers: "List technical specifications including processor, RAM, storage, and connectivity options"
• Apparel & Accessories: "Describe the fabric, fit, care instructions, and style features"
• Home & Garden: "Describe dimensions, materials, assembly requirements, and warranty details"

When a product is enriched, the system automatically selects the appropriate prompt:

1. If a category-specific prompt exists for the product's category, it uses that
2. Otherwise, it falls back to the default prompt

This ensures more relevant and accurate data extraction for different product types.

Exporting Products

Export Selected Products

1. Select products using checkboxes
2. Click Export → Export Selected
3. A CSV file downloads with all product data and attributes

Export All Products

1. Click Export → Export All
2. Downloads complete catalog including all enriched attributes

CSV Format

Exported files include:

• Basic product information (SKU, name, brand, etc.)
• All enriched attributes in separate columns
• Timestamps

Enrichment Settings

Configure organization-wide enrichment behavior in Settings → Enrichment Settings:

Ignored Domains

Specify domains to exclude during enrichment (e.g., analytics, ads):

["google-analytics.com", "facebook.com", "doubleclick.net"]

Country/Region

Set your preferred geographic region for better results

Attribute Management

• View all organization attributes
• Enable/disable attributes
• Reorder attributes
• Edit attribute prompts
• Delete unused attributes

Attribute Sets

Create custom attribute sets to use a subset of attributes during enrichment. See Attribute Sets for details.

• Create sets with specific attributes
• Override prompts per attribute
• Select attribute sets when starting enrichment jobs
• Only organization admins can manage attribute sets

Credits and Costs

How Credits Work

• Credits are consumed each time a product is enriched
• Cost depends on the AI model you choose
• Failed enrichments do not consume credits
• You can monitor credit usage on the Dashboard

Cost Estimation

The platform shows estimated credit cost before starting any enrichment job, helping you plan your usage.

Before You Start

• Check your credit balance before large jobs
• Purchase additional credits if needed from the Credits page
• Set up auto-recharge to avoid interruptions

Best Practices

Preparing Products

1. Add Essential Information First: Fill in SKU, name, and brand before enriching
2. Use Descriptive Names: Better product names lead to better enrichment
3. Include Manufacturer Data: Helps the AI find accurate information
4. Valid URLs: Ensure product pages are accessible (if you're providing URLs)

Choosing the Right Model

• Start with Normal model for most products
• Use Budget for simple products or large batches
• Use Pro for technical products needing detailed specs
• Use Ultra only when accuracy is critical and worth the extra cost

Optimizing Results

1. Test First: Enrich 5-10 products to verify results before bulk processing
2. Batch Wisely: Process 50-100 products at a time for best balance
3. Review and Edit: Always review enriched data and refine as needed
4. Re-enrich When Needed: Update products with new information periodically

Managing Large Catalogs

• Import products via CSV for faster setup
• Use bulk selection to enrich multiple products at once
• Monitor job history to track progress
• Export results periodically for backup

Troubleshooting

Products Not Enriching

• Check Credits: Ensure you have sufficient credit balance
• Verify Product Data: Products need at least a name and SKU
• Check Job Status: View job details for specific error messages
• Internet Connectivity: Enrichment requires stable connection

Poor Quality Results

• Try a Higher Model: Pro or Ultra models provide better accuracy
• Add More Product Info: Include brand, manufacturer
• Manual Editing: Refine attributes after enrichment
• Contact Support: We're here to help optimize your results

Job Stuck or Failed

• Jobs typically process within minutes to hours depending on size
• Check job history for detailed status
• Failed jobs can be retried
• Contact support if issues persist

Next Steps

• Product Validation - Verify enriched data quality
• Manage Credits - Purchase and track credit usage
• API Reference - Automate enrichment workflows
• Dashboard - Monitor enrichment analytics