4 Ways "AI Categories" Eliminates Spreadsheet Mapping

This tool moves beyond simple text matching to understand product intent.

GoDataFeed AI Categories analyzes your source product data—primarily product_type, title, and description—to find the most accurate match in the destination channel's unique taxonomy. This process is fully channel-aware. The AI understands that Google's category for 'throw pillows' (HOME & GARDEN > DECOR > DECORATIVE PILLOWS) is fundamentally different from Amazon's required node (Home & Kitchen > Bedding > Decorative Pillows, Inserts & Covers).

Instead of you manually searching through thousands of nodes in a spreadsheet, you simply select your imported category groups and click "Categorize." The AI reads the underlying SKU data within those groups and returns its best-fit suggestion, populating the mapping field for you.

Pro Tip

For best results, ensure your source product_type field is as detailed as possible. The AI heavily weights this field, so a value like 'Apparel > Shirts > Mens > Polos' provides a much stronger signal than just 'Shirts' and will result in more accurate, granular suggestions.

Technical Insight

The AI doesn't just match keywords; it analyzes the context of your SKUs. If you have a group named "Cases," the AI will check the titles and descriptions. If it sees "iPhone 15" and "Samsung," it will suggest an electronics category. If it sees "Samsonite" and "luggage," it will suggest a travel category.

AI suggestions are a starting point, not a mandate; this workflow keeps you in full control.

GoDataFeed AI Categories is not a black box. It’s a human-in-the-loop system designed for rapid review. Once the AI returns its suggestions, you have three clear actions:

1. Confirm: For suggestions that are obviously correct, you can select them individually or in bulk and click "Confirm." This locks in the mapping.
2. Regenerate: If a suggestion seems off or too broad, click "Regenerate." The AI will re-analyze the product group and provide a different suggestion, often a more granular alternative.
3. Override: When you know the exact category you want, simply click into the "Feed category" cell. This opens the channel's full taxonomy tree, allowing you to manually select the correct node and override the AI's suggestion.

This three-step process allows you to map 90% of your catalog with the "Confirm" button and focus your expertise on the 10% that require manual review.

Pro Tip

Always use the "Regenerate" button on an ambiguous category group before resorting to a manual override. This often finds the right node on the second try and saves you from having to dig through a complex taxonomy tree yourself.

Process Optimization Note

Your review workflow should be:

1. Bulk Confirm all obvious high-confidence matches first.
2. Filter for unmapped products (marked with a red '!' indicator) and use Regenerate on them.
3. Manually Override any remaining outliers or high-priority items that need a specific node.

This tiered approach clears the majority of your catalog in minutes, not hours.

Use 'if-then' logic to manage the specific exceptions within an AI-mapped group.

The AI mapping is excellent for assigning categories at the group level (e.g., all products in your 'Shirts' category). But what about exceptions within that group? Category Rules handle this.

Category Rules are simple 'if-then' statements that run after the AI mapping to refine your data. For example, the AI might correctly map your "Shoes" category to Apparel & Accessories > Shoes. You can then create a rule that says: IF title CONTAINS "Boots" THEN SET google_product_category to Apparel & Accessories > Shoes > Boots.

This granular logic overrides the broader AI mapping for only the SKUs that match your condition. It's the perfect tool for handling accessories, sub-types, or any other predictable exception, ensuring your products land in the most specific category possible.

Pro Tip

Place your most specific rules at the top of the list. The rules engine processes top-down, so a rule for "Waterproof Boots" should run before a more general rule for "Boots" to ensure the correct, most granular assignment.

Real Use Case Scenario

A common challenge is a "Sale" category that contains mixed product types. The AI may struggle to map this group. You can bypass the AI mapping for this group and use rules exclusively.

• Rule 1: IF product_type CONTAINS 'Shirt' THEN map to Apparel > Shirts.
• Rule 2: IF product_type CONTAINS 'Pants' THEN map to Apparel > Pants.

This lets you manage exceptions with repeatable logic without ever touching the source data.

Mappings are theoretical until you compile the feed and see the final SKU-level output.

After all your AI suggestions are confirmed and your rules are built, the final step is to click "Compile." This applies all your logic to every individual SKU in your catalog.

To verify the result, use the "Preview" tool to inspect any product. You can see the exact, final value populated in the channel's category field (e.g., google_product_category or the Amazon feed_product_type). This preview confirms that your logic is working as intended before the feed is sent to the channel. If you run the AI categorization and nothing happens, the most common issue is a missing source mapping. The AI tool needs data to analyze.

Pro Tip

If your category field is blank after compiling, always check your Import Mapping first. The "Category" field in GoDataFeed must be mapped to your source data column (like product_type) before the AI Categories tool can read and analyze it.

Key ROI/Efficiency Metric

Monitor your Google Merchant Center "Diagnostics" tab for "Invalid google_product_category" errors. After implementing AI Categories, this error count should drop to near-zero. This is your primary metric for mapping success and directly impacts your products' eligibility and visibility in Shopping campaigns.