Google’s auto categorization for Shopping is convenient, but once you work with a few thousand SKUs it will misclassify a worrying number of products unless you take control. For serious ecommerce accounts, getting your taxonomy right becomes a core part of feed hygiene, optimization, and ultimately revenue.​

When Google started auto classifying products

Over the last several years, Google has shifted from treating google_product_category as a strictly enforced field to increasingly inferring it when merchants leave it blank. Instead of requiring taxonomy for every item, Google’s systems now lean heavily on title, description, brand, GTIN, and product_type to guess the category.​

This change lowered the barrier to entry for merchants by making feeds easier to launch without perfect taxonomy work upfront. The trade off is that any messy or inconsistent catalog data becomes the training ground for automatic categorization, which can lead to inaccurate or inconsistent category assignments at scale.​

What Google’s product taxonomy actually is

Google’s product taxonomy is a standardized, hierarchical list of product categories that tells Google what you sell and how it relates to other products in the ecosystem. It is structured like a tree, for example:​

• Apparel & Accessories
• Apparel & Accessories > Clothing
• Apparel & Accessories > Clothing > Pants

In your feed, this lives in the google_product_category attribute, which you can provide either as the full text path or as a numeric ID from Google’s taxonomy file. When you do not provide it, Google tries to infer the category based on other attributes using machine learning models trained on product data across Merchant Center.​

On small, tidy catalogs this can work reasonably well, but once there is significant scale or complexity, you often see products from the same family end up in different categories, accessories classified as main products, and whole sets of SKUs dumped into very broad buckets like “Apparel & Accessories”. Because category is a key relevance signal, these misclassifications quickly spill over into performance, reporting, and policy.​

Why having the right taxonomy actually matters

Having accurate, granular google_product_category values is not just a technical nicety; it directly impacts how often you show, where you show, and how effectively you can optimize.​

• It improves query matching and intent. When Google knows an item is “Apparel & Accessories > Shoes > Athletic Shoes > Running Shoes” rather than a generic “Apparel & Accessories”, it can match your products to far more precise, high‑intent searches and avoid wasting impressions on broad, low‑intent queries.​
• It feeds cleaner signals into Smart Bidding and Performance Max. Category is one of many context signals used by automated bidding; if large portions of your catalog are misclassified, models aggregate performance across products that should not be grouped together, making it harder to learn which lines are truly profitable.​
• It unlocks better filtering and reporting. With consistent taxonomy, you can slice performance by product line (for example “Sofas” vs “Chairs”) and make surgical changes to budgets, ROAS targets, and creatives instead of blunt account‑wide adjustments.​
• It helps avoid policy and approval issues. Some policy checks and attribute requirements depend on product category, especially in apparel and sensitive verticals; incorrect taxonomy can trigger misaligned expectations and more disapprovals.​

Real Client example

This Mid size fashion retailer was sending around 5,000 SKUs to Shopping, but almost everything was categorized as “Apparel & Accessories” with no deeper taxonomy. After mapping key families (dresses, jeans, sneakers, jackets) to specific Google product categories and tightening titles, their query mix shifted: generic “clothes” searches declined, while more bottom‑funnel terms like “women’s black midi dress” and “men’s slim fit jeans” increased. CTR and conversion rate improved on those mapped product lines because Google finally understood the difference between dresses, sneakers, and jackets instead of treating them as one amorphous blob.

Why auto categorization breaks on larger catalogs

Google assigning one for you sounds appealing, but real catalogs are messy. Once you cross a few thousand SKUs, you typically face inconsistent titles and product_types, similar language used for different items, and legacy site categories that no longer match how products are actually merchandised. Automatic systems read this noisy data and do their best, but the outcome is often similar products scattered across categories, many SKUs collapsed into generic top level buckets, and odd one offs where a single keyword pushes an item into a completely wrong vertical.​

For advertisers who care about efficiency and scale, “let Google figure it out” stops being viable. The solution is not to abandon automation, but to give it a clean, reliable taxonomy foundation.

Real Client Example

This MegaMall aggregates thousands of SKUs from multiple brands and suppliers, each with their own naming conventions. With auto categorization left on autopilot, Google inferred categories such as fashion belts being classified under jewelry because of words like “chain”, decorative cushions split between “Bedding” and “Home Décor”, and identical phone cases scattered between “Electronics Accessories” and “Cases & Bags”. Shopping reports by category became unreliable, and PMax grouped dissimilar products together, which confused bidding and creative signals. After building a central taxonomy mapping and overriding google_product_category, performance stabilized and reporting began to reflect reality.

A practical approach for getting taxonomy right at scale

Manually editing 3,000 products is not realistic, so the goal is to manage taxonomy at the level of product families instead of individual SKUs.​

• Create a clean internal structure for your products using product_type or your platform’s category path as the “source of truth”.​
• Group SKUs into logical families based on that structure or recurring title patterns.
• Map each internal group to the closest matching Google product category, using either full paths or numeric IDs.​
• Implement rules in your feed process so that when product_type contains specific patterns (“Pants”, “Running Shoes”, “Sofas”), the correct google_product_category is applied without manual edits.​
• Prioritize high‑impact categories and then iterate outwards, auditing taxonomy periodically in Merchant Center to catch anomalies and new product lines.​

When this is in place, Google’s automation has better signals, your reporting becomes clearer, and your optimization work becomes more targeted. Taxonomy stops being an invisible drag on performance and starts acting as a quiet multiplier on everything you do in Shopping and Performance Max.