Guidelines for Enriching Product Attributes

AI-powered enrichment finds relevant product information, but maintaining consistency requires well-defined standards. While we custom-train our AI to align with your company’s needs, you can also take control by defining attribute guidelines yourself. Here’s how you can refine enrichment results for better accuracy and consistency.

Giving Definitions to Each Attribute

AI relies on clear definitions to accurately enrich product data. Without them, generic field names like "Size" or "Material" can lead to inconsistent or incorrect enrichment. Clearly defining each attribute ensures AI understands the intended meaning and applies the correct data.

Example Attribute Definitions

• Size (Apparel): The numerical or letter-based measurement of a garment, typically based on regional sizing standards (e.g., US, EU, UK).
• Size (Packaging): The physical dimensions of a product’s packaging, expressed in length × width × height (L × W × H).
• Finish: The surface treatment or texture of a product, such as glossy, matte, or brushed metal.
• Fit: The way a garment conforms to the w body, such as slim fit, regular fit, or oversized.

By providing clear, distinct definitions for each attribute, you help the AI distinguish between similar fields and apply the correct enrichment.

Formatting Text-Based Attributes

In our previous article on Managing Fields and Properties, we discussed different attribute structures. For short and long text fields, you can refine enriched values by defining formatting rules in the guidelines.

Example Formatting Guidelines

• Voltage Format: Always display voltage ranges as 230V - 250V , ensuring proper spacing and unit placement.
• Product Titles: Use Title Case for product names (e.g., "Men’s Running Shoes").
• Bullet Points for Features: Ensure long descriptions use bullet points for key features instead of paragraphs.

These guidelines help AI generate standardized, readable product attributes.

Handling Single and Multi-tags Attributes

Predefined lists improve accuracy by restricting AI-generated values to a set of approved terms. For instance, instead of unpredictable variations like "Crimson" or "Scarlet," you might want all red-colored bags to be labeled simply as "Red."

Defining a List of Values

To set predefined values:

1. Navigate to a Single tag or Multi tags cell. Double-click on the cell.
2. Enter the approved list of values (e.g., Red, Blue, Green ).

3. Save to enforce consistent enrichment.

   

There are 2 options

• Strict Matching (default): AI only assigns values from the predefined list.
• AI-Suggested Values: AI can suggest new options when an exact match isn’t available.

Free-Form Dynamic Tags

Unlike predefined lists, free-form tags allow flexible, descriptive enrichment. This is particularly useful for:

• SEO and discoverability (e.g., search keywords).
• Contextual product details (e.g., "suitable for a day out").
• Customer targeting (e.g., "ideal for young professionals").

To enable free-form tagging:

1. Navigate to a Multi-tags field.
2. Change the property from “List of Values” to “Free Form”.

You can use guidelines to guide the tags the AI generates.

Example guidelines

• Always use lowercase (e.g., "lightweight, breathable").
• Include relevant occasions (e.g., "suitable for a summer vacation").
• Specify product style (e.g., "modern, minimalist").
• Identify the target audience (e.g., "perfect for outdoor enthusiasts").
• Include trend-based descriptors (e.g., "2024 fashion must-have").

By defining attributes, formatting text, managing selections, and refining tags, you can shape AI enrichment for better, more consistent product data.

If you need further support, please reach out to support@hypotenuse.ai and a member of our team would be more than happy to assist you!