How to find the right Google product category (with ID)

What google_product_category actually is

google_product_category is a Merchant Center attribute that holds one value from Google’s own fixed taxonomy — a tree of 5,595 categories, from Animals & Pet Supplies at the top down to leaves like Apparel & Accessories > Shoes > Athletic Shoes. You do not invent the values. You pick one that already exists.

Two things about it surprise people. The first: it is an optional attribute. Google’s documentation is explicit that all products are automatically assigned a product category from that taxonomy, whether or not you submit one. A product is not disapproved for lacking a category, and Google publishes no “missing Google product category” fix article — though it does publish one for missing brand.

The second: submitting an invalid value is worse than submitting nothing. Google’s guidance on invalid categories says that if the attribute is not provided it will use internal signals, but that invalid values will be rejected and may prevent your products from appearing. So a guessed string is a real risk, and an exact taxonomy value is not.

A category that is wrong and specific is worse than one that is right and broad.

When the category you pick actually changes something

This is the question worth answering before you spend an afternoon on a spreadsheet. Google’s documentation describes the attribute as something that can override its automatic categorisation in specific cases, and names them:

• Enforcement of category-specific attribute requirements. Some categories bring their own required attributes — Apparel & Accessories, Mobile Phones and Software all do. Your category determines which rules apply to the rest of your feed.
• Google Ads campaign targeting. Product groups and bidding in Shopping campaigns can be organised by category, so the value decides what you can target.
• Alcohol. Alcoholic beverages have to sit in specific categories.

There are two more concrete cases in the product data specification. A product using subscription_cost must have google_product_category present — that is the one place the attribute is genuinely mandatory. And a handful of product types must use particular IDs: a gift card is ID 53, a phone sold with a contract has a short list of allowed IDs including 267.

Outside those cases, the strongest practical reason is that the column exists in other systems too. Meta’s catalog documentation describes google_product_category as representing the item according to the Google’s product taxonomy and links to the same taxonomy file. Microsoft accepts the Google-formatted attribute and defines its own field as equivalent — though it publishes its own taxonomy, so that is not the same as saying Bing uses Google’s. Feed plugins across WooCommerce and Shopify all have a column for it. Fill it once, correctly, and you have filled it for several channels.

Get the taxonomy file

Google publishes the full list in two places:

• taxonomy-with-ids.en-US.txt — one category per line, ID first, then the full path. This is the one to work from.
• taxonomy-with-ids.en-US.xls — the same thing for people who want a spreadsheet.

Both live under google.com/basepages/producttype/. The taxonomy is described as continuously evolving, and Google will accept any version of the categories published after August 2011, translating older values to the current version. That is forgiving, but it also means a category you cached two years ago may have moved.

Pick the most specific category that is still true

That sentence is the whole method, and both halves matter. More specific is better, because a deeper category is matched against more specific shopping queries and carries more precise attribute rules. But a category that is wrong and specific is worse than one that is right and broad.

Work down the tree rather than searching for a leaf name:

1. Pick the top-level branch. There are 21, and this is usually obvious. If it is not — a cat window perch could be Animals & Pet Supplies or Furniture — decide what the product is, not who buys it.
2. At each level, read all the siblings before choosing. Names repeat across the tree: “Accessories” appears many times, and the parent is what distinguishes them.
3. Stop when the next level down would require information the product does not actually have. Google accepts interior categories, and Apparel & Accessories > Shoes is the honest answer for a listing that does not say what kind of shoe.

A shortcut that is not cheating: our Category Finder walks the tree the same way and returns the ID, the confidence, and the runner-up categories it weighed against. The runners-up are the useful part — they tell you which decision was close, which is exactly where a human should look.

Submit the ID, not the path

Google accepts either the ID or the full path, but not both, and the product data specification says outright that it is recommended to use the category ID.

Use the ID. A path is a string that has to match exactly, including the > separators and the ampersands, and it changes when Google renames a category. The ID does not. If you are filling this in by hand across hundreds of rows, the ID is also four characters instead of sixty.

Doing it for a whole catalog

Category by category is fine for twenty products. At a thousand it stops being work you can do carefully, which is the real problem — a tired person assigning categories makes systematic mistakes, not random ones.

Two things make it tractable. First, deduplicate: a catalog with one t-shirt in six sizes and four colours has one category decision, not twenty-four. Group by product before you start. Second, separate the confident rows from the uncertain ones and only read the uncertain ones. Our Bulk Category Mapper does both, and outputs the low-confidence rows as their own file with the runner-up categories attached, so the manual review is a short list rather than the whole catalog.

Four mistakes that cost real time

• Confusing it with product_type. They are different attributes doing different jobs. The difference, with what each one is for.
• Typing a path from memory. “Apparel > Shoes” is not a valid value; the branch is “Apparel & Accessories”. Copy, do not retype.
• Assuming your value always wins. It overrides Google’s automatic categorisation in specific cases, not universally.
• Re-importing a CSV with the column blank. On Shopify, a blank non-required column in an import overwrites the existing value with blank. A partial re-import can wipe categories you already set.