Automatically Categorize Your Products To Google Product Taxonomy in 4 Steps

Matt Payne
September 6, 2022

Ecommerce brands and marketplaces use product taxonomy structures to organize their products so customers can find exactly what they are looking for in the least amount of time and clicks. The hierarchy of products makes it easier for customers to refine what they want and easily filter out what they don’t. 

What is Google Product Taxonomy?

google product taxonomy in neo4j
Google shopping categories in Neo4j

Google product taxonomy is an ecommerce taxonomy tree with 5,595 unique categories used by Google Shopping, Bing, and Shopify. This tree is used by ecommerce stores to either internally structure product hierarchy or to use Google Shopping. 

Why Use Google Shopping Product Categories?

The most direct reason for using google product taxonomy is to meet requirements for adding your products to your Google Shopping feed. Google will automatically categorize your product based on titles, descriptions, price, brand, and GTIN, but you can override the assigned category with the [google_product_category] attribute. Having your product correctly categorized with google product taxonomy makes it easier for shoppers. Google has a nice guide here for the optional google product category tag. These categories also make it much easier to compare Google ads campaigns.

Another key reason for using google product categories is the increase you see in SEO, traffic, and conversions through the improved product catalog structure. Providing a concrete taxonomy structure to your available products makes it easier for potential customers to find what they’re looking for in as few searches as possible. This is a huge part of conversions as 79% of your visitors will head to a competitor site for the same product if they can’t easily find what they want to purchase. It shouldn’t be a surprise then that Forrester found that poorly structured sites sell 50% less than organized sites with solid product taxonomy.

The improved product catalog structure benefits go off site as well. Well structured sites with on-topic product pages lead to more organic traffic through search engines. If organic traffic is one of your goals then moving up in Google has to be priority as only 0.78% of Google searchers click to the second page, and the number 1 ranking site for keywords gets 49% of all search traffic.

How Granular Should Your Store Be With Google Shopping Taxonomy?

The level of granularity you want to use with Google shopping taxonomy should follow a few parameters that help Google Shopping and your own on-site conversions.

Google Shopping feed granularity

Some product categories in Google product taxonomy have a higher level of granularity than others and therefore have deeper nodes in the tree. When manually overriding the assigned category from Google Shopping you should make sure specific information in your product data supports the new category. Oftentimes the previous category placement had to do with specific information not being on-topic enough to be given a more granular category. It’s also worth noting some categories do not allow for manual override.

google product taxonomy example data
Fitting product information to the correct google product category is important for your Google Shopping feed  

You’ll also want to assess your product data and the granular category you think you belong in relative to the other available categories at the same level. 

Internal taxonomy structure granularity

When leveraging the Google product taxonomy as an internal taxonomy structure it’s important to understand the categories relative to your products. The level of granularity you decide on for a given product should depend on the other products in the same category and how correlated they are. Ai based product matching tools allow you to compare various products for similarity. Tools like this make it easy to understand how similar your products are in the same category and decide if you should move to a more granular category.

Unsure of how granular your category should be for a specific product or what category is the best fit? Let’s look at how artificial intelligence can be used to automate the process of categorizing products to Google product taxonomy.

Automatically Fit Your Products To Google Product Taxonomy In 4 Steps

Artificial intelligence has made it easy to automatically categorize products into the Google product taxonomy across the nearly 6000 unique categories available in the most updated tree. No need for tricky transformation rules or keyword matching as our automated product categorization software has learned the relationship between the Google Shopping categories and product data with over 95% accuracy. 

You can go from product data and taxonomy tree to structured data in a matter of minutes with Pumice.ai

Step 1. Gather Product Data or Connect To Our API

We first want to handle collecting the data and preparing it for processing through our product categorization endpoints. Pumice.ai operations require just product title and description but can leverage other fields like Price, GTIN, and SKU for increased accuracy.

raw api connection in Pumice.ai

You can attach your data to the ai operations through a csv upload and field mapping in the dashboard or directly connect to the API. Raw API connection allows you to add these ai endpoints to a part of your workflow automatically to boost the level of automation in your business process. Enterprise customers can leverage custom integrations to automatically connect to ecommerce platforms such as Shopify and Woocommerce.

Step 2. Select Google Shopping Taxonomy In Pumice.ai

dynamic api in pumice.ai for a batch run

The Google Shopping taxonomy tree is already a built in option in any of our product categorization endpoints. All of our models have been trained with this tree as a part of their training. Simply select the “google_product_taxonomy.txt” option in the endpoint that best fits your use case. Let’s figure out which endpoint is the best fit for you.

Step 3. Automatically Fit Your Products To Google Shopping Taxonomy Tree

product data and product catalog data creates pumice.ai enhanced data

Pumice.ai offers a number of different API endpoints that can be used to fit product data to the Google Shopping taxonomy tree. These endpoints vary in terms of required data at runtime and the nature as to how they categorize products. 

Dynamic API to fit products to categories

pumice.ai dashboard

The dynamic API allows you to categorize your product data to a taxonomy tree or category list dynamically at runtime. This means the model does not require training specifically for your categories and offers the most flexibility with very little tuning for each tree. The key benefit of this endpoint is the speed at which you can use multiple trees with the same model without training specific to your categories.

Trained API to fit products to categories

The trained API is a product categorization endpoint trained specifically on your taxonomy tree and product data. This endpoint contains ML models that only support the taxonomy structure they are trained on and will best fit products specific to your tree. The main benefit to this endpoint over the dynamic one is that it tends to have a higher accuracy for use cases where the product categories have little correlation to product information. The downside being that adjusting your tree or categories requires retraining.

Product similarity API for dynamic categorization

product similarity in pumice.ai

Our product similarity API allows you to compare the similarity of two product records. This endpoint can be setup to allow you to best fit new products on the fly based on real time category to product matches in your catalog. While this requires more setup and a deep product catalog the results can be more inline with often changing product catalogs.

Step 4. Integrate New Data Into Product Information Management (PIM) System

automated business workflow for ecommerce companies leveraging pumice.ai

Once you’ve ran the endpoints and generated your product taxonomy matches you can leverage this data by passing it back into your PIM. The data from the endpoints is returned in the same format you passed it in as unless you have hooked up a custom integration. These custom integrations allow you to automate more of this product information pipeline and remove more manual labor from product categorization. Our most popular integrations are:

- Shopify

- Woocommerce

- Google Sheets

- Jasper PIM

- Salsify

- Sales Layer PIM

- Your centralized product data storage system

Once you’ve generated your data you’re done! Most of our API endpoints are one click runs so you can sit back and relax while our Ai automatically categorizes your products to Google product taxonomy. 

Start automating your product categorization today

Pumice.ai is a PIM enhancement platform built to allow you to leverage our proven Ai/ML tools to automate product information tasks at scale. You can leverage custom ai models and integrations in our dashboard with the development team at Width.ai. Contact us today to learn more about how you can start automating your product management tasks and start seeing increases in conversions and reductions in costs.

pumice.ai logo

List of Google Product Categories

Here’s a list of the Google Product Categories as provided by google.