🛍️ Product Feeds: Best Practices

By following these guidelines, the chatbot can retrieve the correct product information, minimize hallucinations, and provide relevant recommendations.

Ideally, use the Google Merchant XML format which we optimize on the go.

		<rss version="2.0" xmlns:g="http://base.google.com/ns/1.0">

If you have a custom feed, follow the best practices when structuring your product feeds.

🔝 1. Prioritize Important Data

The bot searches content from top to bottom. Sort the data from the most important to the least important:

• 🏷️ Product ID
• 📢 Title
• 💰 Price
• 🎯 Sale Price
• ... some less important data

For example, if you place crucial data like price at the end of the product item, the bot might not see it and could generate inaccurate responses.

📏 2. Maintain Data Consistency & Structure

• Use a well-defined XML schema with clear and consistent field names.
• Ensure each product has a unique, persistent ID to avoid confusion in updates.
• Use separate fields for each data type (e.g., title, description, price, brand, category, images, stock).

📝 3. Optimize Product Titles & Descriptions

• ✂️ Keep Entries Concise
  • Shorter content is better. Large entries cost more and slow down responses.
• Title Format
• Use a structured format:
  [Brand] [Product Name] [Key Attributes] (e.g., size, color, model)
  • Example:
    Nike Air Max 270 - Men's Running Shoes, Black/White, Size 42
• 📌 Description Best Practices
  • Start with a concise overview.
  • List key features in bullet points for better readability.
  • Avoid generic marketing fluff—focus on specifications and unique selling points.

🎯 4. Improve Categorization & Attributes

• Use standardized categories (e.g., Google Taxonomy) to improve searchability.
• Provide structured attributes (e.g., material, compatibility, dimensions).
• Include synonyms for product types (e.g., "sneakers" vs. "running shoes") in metadata fields to help GPT match queries.

🚧

Common Mistakes to Avoid

❌ Do not put important data at the end of the product item.

❌ Long, unstructured descriptions that mix marketing and technical data.

❌ Overloaded product titles with unnecessary details.

❌ Duplicate products with minor variations—use variation fields properly (e.g., sizes in a separate attribute).

❌ Missing or incorrect identifiers (GTIN, SKU, MPN, etc.).