Google has updated its Google Ads help documentation to state explicitly that it automatically assigns products to categories using a continuously evolving product taxonomy. This wasn't news to anyone managing Merchant Centre feeds - that documentation has carried similar wording for years. But the fact that it now sits on the Google Ads side of the documentation matters. It signals that category assignment is no longer just a feed hygiene issue. It is part of how Google Ads itself understands, organises, and serves your products.
For advertisers who have been treating Google product categories as a static administrative field, this is a prompt to think again. If the taxonomy evolves and Google is doing the assigning, then the category your product sits in today may not be where it sits in six months. That has consequences for auction eligibility, campaign performance, and how your products behave inside Performance Max.
What 'Continuously Evolving' Actually Implies
The phrase 'continuously evolving product taxonomy' is doing a lot of work here. It means Google reserves the right to reclassify products as categories are added, split, or restructured. A product sitting comfortably in a broad category today could end up in a more specific sub-category as Google refines its taxonomy. That shift affects which queries the product is considered eligible to match, and in some cases, which competitors it is placed alongside.
This is not a hypothetical concern. Google's product taxonomy already runs to thousands of categories across dozens of verticals. When categories are refined - say, a general 'Home & Garden > Furniture' branch gets more granular sub-nodes - products that were broadly categorised can end up in unexpected places if the automatic assignment logic re-evaluates them. The feed data you submit, including product titles, descriptions, and custom labels, feeds into that assignment logic. Weak or generic feed attributes give the algorithm less signal and increase the chance of a poor category fit.
The Merchant Centre Gap Has Now Closed - Partly
Merchant Centre has documented automatic category assignment for several years. Experienced feed managers have long understood that submitting a google_product_category value in the feed can influence but does not guarantee the final assigned category. Google's system weighs your submitted value alongside the product data itself and makes its own determination. What has changed is that this behaviour is now formally acknowledged on the Google Ads product category insights documentation.
This matters because many Google Ads practitioners - particularly those who are not deep in feed management - may have assumed category assignment was a Merchant Centre concern that didn't affect their campaign decisions. That assumption is now explicitly incorrect. The category insights data visible inside Google Ads is generated by the same automatic assignment process. If you are using that data to structure asset groups, review product performance, or inform bidding decisions, you need to understand that the underlying category labels are dynamic.
How This Affects Performance Max and Asset Group Structure
Performance Max uses product category data as one of the signals it uses to understand inventory and serve relevant ads. When you segment asset groups by product type or category, you are working with Google's taxonomy - not your own internal classification. If Google reassigns a batch of your products to a different category, the segmentation logic you built your asset groups around can quietly shift beneath you. The structure looks the same in the interface, but the product population inside each group may have changed.
This is particularly relevant for retailers with large catalogues or seasonal product ranges that straddle multiple categories. A product that reads clearly as 'Apparel & Accessories > Clothing > Outerwear' in winter might be reclassified as something adjacent when Google updates how it interprets product signals. For PMax campaigns where budget and bidding decisions are tied to specific asset group performance, an unexpected reclassification can skew the data you are relying on to make those decisions.
The practical implication is that category-based segmentation in PMax should be treated as a starting point, not a permanent structure. Building custom labels into your feed - based on margin, priority tier, or product line - gives you a stable segmentation layer that you control and that won't shift with Google's taxonomy. Product type attributes serve a similar function. These are fields where your data wins over Google's interpretation.
Feed Quality as a Campaign Performance Input
The more complete and specific your product data, the better chance you have of influencing where Google places your products in its taxonomy. Title structure matters here. A title that front-loads the most specific product attributes - brand, product type, key differentiator, size or variant - gives Google's classification system more to work with than a vague or brand-led title that buries the product type. The same applies to product descriptions, which should reinforce category-relevant terms rather than read like marketing copy.
Google does allow advertisers to submit a google_product_category value in the feed, and doing so is still worth the effort. Even if Google may override your submission, providing an accurate category value establishes a clear signal. Submitting no value, or a broad top-level category when a more specific one exists, is the equivalent of giving the algorithm an open field and hoping it finds the right answer. With a continuously evolving taxonomy, that approach carries more risk than it used to.
Monitoring Category Assignment as an Ongoing Task
Product category insights in Google Ads give you visibility into how Google is categorising your inventory. This report is worth checking periodically - not just at launch - precisely because the taxonomy evolves. If you see products appearing in categories that don't match your expectations, that is a signal to investigate the underlying feed data for those products. The issue is usually a weak product title, a missing product type, or an absent google_product_category value.
For agencies managing Shopping or PMax on behalf of clients, this is also worth raising in account reviews. Category assignment is not something most clients think about, but a misclassified product range can affect impression share, auction eligibility, and ultimately cost per acquisition. Framing feed quality as a direct input to paid performance - rather than a back-office feed management task - tends to get the attention it deserves.
The broader point is straightforward: Google has formalised that it decides how your products are categorised, and it reserves the right to keep refining that decision. Working with that reality - by investing in feed quality, using stable custom labels for campaign segmentation, and monitoring category insights data - is more reliable than assuming the taxonomy will stay still.