Skip to main content

Text Extraction enrichment

Text Extraction enrichment is used to extract structured product data from unstructured text sources.

S
Written by SKULaunch Support
Updated over 4 months ago

Text Extraction enrichment is used to extract structured product data from unstructured text input.

This agent reads pasted or existing text and identifies values that can be mapped to the attributes you have selected. It is designed for data enrichment, not content generation.

When to use Text Extraction enrichment

Text Extraction enrichment is a good choice when:
• Product information exists as raw text
• Specifications are embedded within paragraphs
• Data has been copied from documents, emails, or websites

It is commonly used when information is available, but not already structured into attributes.

What Text Extraction enrichment is good at

Text Extraction enrichment is commonly used to populate:
• Technical specifications
• Dimensions and measurements
• Performance ratings
• Materials and standards

It works well when specifications are clearly written, even if they are not formatted consistently.

What it is not good at

Text Extraction enrichment should be avoided when:
• Information only exists in images or scans
• Data is locked inside files that cannot be copied as text
• Text covers multiple different products at once

In these cases, another enrichment agent will produce better results.

What Text Extraction enrichment uses as input

This agent relies on:
• Pasted text
• Existing product descriptions
• Copied specifications from documents or web pages

The cleaner and more focused the text, the better the results.

How results should be treated

Values returned by Text Extraction enrichment are:
• Suggested attribute values
• Always reviewable and editable
• Dependent on the clarity of the input text

You should review enriched values before approving them for use.

Tips for better results

• Paste text for a single product at a time
• Remove irrelevant marketing content where possible
• Start with a small number of attributes
• Use clear attribute definitions and AI guidelines

Did this answer your question?