Skip to main content
Product enrichment is the process of automatically generating structured attributes and content for your products using AI. Instead of manually writing descriptions, identifying colors, or filling in product properties, Cernel’s AI does it for you—intelligently and at scale.

​
What Is Product Enrichment?

Enrichment transforms incomplete or basic product data into rich, detailed information: Before Enrichment: 
Title: Blue Cotton T-Shirt
Description: (empty)
Color: (not set)
Material: (not set)
After Enrichment: 
Title: Blue Cotton T-Shirt
Description: "This classic crew neck t-shirt combines comfort and style..."
Color: Blue
Material: Cotton
Care Instructions: "Machine wash cold, tumble dry low..."
Enrichment doesn’t just fill in templates—it uses AI to understand your product context and generate appropriate, natural content.

​
Types of Enrichment

Cernel can generate three categories of product attributes:

​
1. Descriptive Content

Free-text content that describes and markets your products:

Product Description

Full description (200-500 words) highlighting features, benefits, and use cases

Short Description

Brief summary (50-100 words) for product cards and previews

Features & Benefits

Bullet-point lists of key product features

Care Instructions

How to clean, maintain, or use the product
Example Use Cases:
  • Clothing: Fit, fabric feel, styling suggestions
  • Electronics: Technical capabilities, use cases, compatibility
  • Home goods: Dimensions, materials, room placement ideas

​
2. Product Attributes

Structured properties used for filtering and search:

Single-Select

One value from a listColor, Size, Gender, Brand

Multi-Select

Multiple values from a listColors (for multi-color items), Features, Compatible With

Numeric

Numeric valuesWeight, Dimensions, Capacity
Common Attributes:
  • Apparel: Color, Size, Material, Gender, Age Group, Pattern, Style, Occasion
  • Electronics: Brand, Model, Connectivity, Power Source, Compatibility
  • Home Goods: Material, Dimensions, Room, Style, Color Family
Structured attributes enable product filtering in your store (“Show me Red Dresses under $100”) and improve SEO for attribute-based searches.

​
3. SEO Content

Search-engine-optimized content for better organic visibility:

Meta Title

Search engine title tag (max 55 characters)

Meta Description

Search result snippet (max 160 characters)

Image Alt Text

Accessibility and SEO text for product images
Learn more about SEO optimization →

​
How Enrichment Works

Cernel’s enrichment process combines multiple AI techniques:
1

Context Gathering

Cernel analyzes:
  • Product title, description, and existing attributes
  • Primary collection (category context)
  • Related products in the same collection
  • Product images (visual analysis)
2

Prompt Compilation

For each attribute, Cernel compiles a custom prompt:
  • Base instructions for the attribute type
  • Collection-specific overrides (if configured)
  • Product-specific data insertion (title, brand, etc.)
  • Constraints (character limits, format requirements)
3

AI Generation

The prompt is sent to the AI model (Gemini, GPT, etc.) which generates content based on understanding of the product context
4

Validation

Generated content is validated against constraints:
  • Character limits
  • Required format (HTML, plain text, etc.)
  • Value types (single vs. multi-select)
5

Review & Apply

You review the AI-generated content, make edits if needed, and apply it back to your store

​
Starting an Enrichment

You can start enrichments from three locations:
  • Products Page: Select specific products and choose attributes to enrich
  • Collections: Enrich all products in a collection at once for consistency
  • Individual Product: Enrich specific attributes for a single product
For detailed step-by-step instructions on running your first enrichment, see the First Enrichment Tutorial.

​
Enrichment Quality Factors

The quality of AI-generated content depends on several factors:

​
1. Source Data Quality

Better input = better output:
  • Products with detailed titles perform better than vague ones (“Men’s Cotton Dress Shirt” > “Shirt”)
  • Existing descriptions provide context even if they’re being replaced
  • High-quality images enable visual attribute extraction
  • Correct primary collection assignment gives proper context
Garbage in, garbage out: If your source product data is incomplete, inconsistent, or low-quality, AI output will reflect those issues.

​
2. Primary Collection Assignment

The AI uses your product’s primary collection to understand context:
  • A dress in “Formal Wear” gets different language than one in “Beachwear”
  • Electronics in “Gaming” emphasize performance, while “Business Laptops” emphasize productivity
  • Kids’ products use simpler language and emphasize safety
Always ensure products have the correct primary collection before enriching.

​
3. Prompt Configuration

Customize prompts to control AI behavior:
  • Tone: Formal, casual, playful, technical
  • Data sources: Which product fields to reference
  • Constraints: Length limits, required keywords, format
  • Layout: Structure for HTML content
Learn how to customize prompts →

​
4. Model Selection

Different AI models have different strengths:
  • Gemini: Fast, cost-effective, good for structured attributes
  • GPT-4: High quality for complex descriptions and nuanced content
  • Claude: Strong reasoning, good for creative content
Most users should start with the default model and only switch if quality isn’t meeting expectations.

​
Reviewing Enriched Content

After enrichment completes, always review AI-generated content before applying it. Evaluate content across these dimensions:
  • Accuracy: Does the content accurately describe the product? Check that attributes match reality (color, material, features)
  • Consistency: Is the tone and style consistent with your brand?
  • Completeness: Does the content cover important features without gaps?
  • Grammar & Readability: Is the writing clear, grammatically correct, and easy to read?
  • SEO Quality: For meta tags, are they compelling, keyword-rich, and within character limits?
Review a sample of enriched products carefully. If quality is consistently good, you can batch-apply with confidence. For detailed review workflow instructions, see the Review Workflow Guide.

​
Best Practices

Enrich 5-10 products first, review quality, refine prompts, then scale to your full catalog.
Group similar products (by collection) for consistent tone and style. Don’t mix luxury items with budget items in the same enrichment.
Start with best-sellers, hero products, or new arrivals where quality matters most.
Review a representative sample to assess quality. If the sample is good, batch-apply the rest.
If enrichment quality isn’t meeting expectations, refine your prompts rather than manually editing every product.
Read the AI’s reasoning to understand its logic. This helps you spot data quality issues and improve source data.

​
Common Enrichment Scenarios

​
Scenario 1: New Product Launch

Goal: Quickly create complete product content for 50 new products. Workflow:
  1. Import products from your platform
  2. Assign to correct collections
  3. Enrich all attributes (descriptions, SEO, structured attributes)
  4. Review content
  5. Publish to store
Time Saved: ~10-15 hours of manual writing.

​
Scenario 2: Existing Catalog Enhancement

Goal: Add structured attributes (color, material, etc.) to 5,000 existing products for better filtering. Workflow:
  1. Create groups for testing (e.g., “Sample 10 Products”)
  2. Test enrichment and review quality
  3. Enrich by collection in batches of 100-500
  4. Review and apply progressively
Time Saved: Weeks of manual categorization.

​
Scenario 3: SEO Improvement

Goal: Add meta titles and descriptions to entire catalog for better search rankings. Workflow:
  1. Enrich meta title and meta description for all products
  2. Prioritize high-traffic products for manual review
  3. Batch-apply for long-tail products
  4. Monitor search performance improvements
Time Saved: Days of copywriting work.

​
What’s Next?

SEO Optimization

Deep dive into SEO content generation

Bulk Operations

Enrich hundreds or thousands of products at once

Attribute Prompts

Customize how AI generates each attribute

Review Workflow

Learn the step-by-step review process
Next: Learn about SEO Optimization to drive organic traffic with AI-generated content.