Merchant Center Automatic Review: The Role of AI

The Merchant Center automatic review has always existed: every product uploaded to the feed has always been subject to automatic verification before being approved or disapproved. What has changed radically over recent years is the technology behind that verification. The system that operated on static rules, "if field X is empty, disapprove", has given way to a system based on Merchant Center artificial intelligence, capable of reasoning by context, semantically comparing feed and landing page, and assigning a continuous quality score instead of a simple approved/not-approved decision.

This change has had significant operational consequences. The Merchant Center automatic review has become faster for standard products and less predictable for products with ambiguous attributes. Merchant Center artificial intelligence detects inconsistencies that previously went unnoticed, but it also produces disapprovals that look unjustified to the human eye.

This article analyses what has changed, how the new system works, and how to adapt, always distinguishing between what Google officially documents and what emerges from hands-on experience on real accounts.

What Has Changed in the Merchant Center Automatic Review

The previous Merchant Center automatic review system operated primarily on rule-based checks: mandatory attributes present or absent, price format correct or incorrect, landing page URL reachable or not. It was a binary, predictable system: knowing the rules meant being able to avoid almost all disapprovals.

Merchant Center artificial intelligence has introduced three structural changes that official documentation only describes in part.

1. From binary decision to continuous scoring

The old system produced a binary decision: approved or disapproved. Merchant Center artificial intelligence instead assigns a continuous quality score to each product, an internal value Google does not expose directly in the interface but which determines both formal approval and visibility in the auction.

A product with a low score can appear technically approved but with reduced visibility, the so-called "silent disapproval" that has become far more frequent with the AI-based Merchant Center automatic review.

2. Semantic analysis instead of pattern matching

The old system compared feed fields against predefined rules. Merchant Center artificial intelligence analyses the meaning of the content: it compares the product title against the declared category, verifies semantic coherence between the feed description and the landing page content, and evaluates whether the images actually show the declared product.

3. Learning from account history

Merchant Center artificial intelligence takes account history into consideration when evaluating individual products. Accounts with a strong quality track record tend to receive a faster and less stringent Merchant Center automatic review on borderline individual attributes.

How Merchant Center Artificial Intelligence Works During Feed Verification

Image analysis with computer vision

The Merchant Center automatic review uses computer vision models to analyse product images. This allows it to detect: overlaid text on the image (promotional banners, watermarks, prices), the presence of multiple products in the main image when not permitted, images that do not actually show the declared product, and, with growing accuracy, low-quality or blurred images.

Semantic comparison: feed vs landing page

Merchant Center artificial intelligence verifies overall semantic coherence: does the product title in the feed match the product heading on the page? Are the declared attributes (colour, material, specifications) visible on the landing page? Is the product actually purchasable and not just informational?

Automatic category classification

Merchant Center artificial intelligence can infer the correct product category from the title, description, and images, independently of the category declared in the google_product_category field. When the category inferred by AI diverges significantly from the one declared, the Merchant Center automatic review may apply the requirements of the inferred category, causing disapprovals for attributes that would not have been mandatory under the declared category.

New Disapproval Patterns Introduced by AI

The introduction of Merchant Center artificial intelligence in the Merchant Center automatic review has generated disapproval patterns the old system did not produce.

Disapprovals for "context inconsistency"

The Merchant Center automatic review can disapprove a product for inconsistency between the site's context (brand, predominant category, buyer type) and the characteristics of the individual product in the feed. A predominantly B2C site uploading clearly industrial products, or a women's clothing site uploading workwear equipment, can receive disapprovals not for specific attribute violations but for context inconsistency, a type of detection impossible for a rule-based system.

Progressive degradation before formal disapproval

The most significant change introduced by Merchant Center artificial intelligence in the Merchant Center automatic review is the progressive degradation of visibility before a formal disapproval occurs. The system lowers the product's score over time, reducing its visibility in auctions, without producing an explicit error message in the Diagnostics section.

False positives on images with complex backgrounds

The computer vision models used in the Merchant Center automatic review generate a share of false positives on images with complex backgrounds, pronounced shadows, or reflections that the model incorrectly interprets as overlaid text or non-compliant graphic elements.

How to Adapt Feed Management to the AI-Based Automatic Review

Optimise for semantic coherence, not just formal requirements

With Merchant Center artificial intelligence, filling in mandatory attributes is no longer sufficient: they must be coherent with each other and with the landing page content. A title that describes a different product from the one shown in the image, or a category that does not match the real product, now generates negative signals in the Merchant Center automatic review that the old system would not have detected.

Monitor feed score over time, not just formal disapprovals

The "Data quality issues" section in Merchant Center is the most direct proxy of the internal score assigned by Merchant Center artificial intelligence. It should be monitored weekly alongside the "Disapproved items" tab, because the issues with the greatest impact on visibility often appear only in the first tab, without producing formal disapprovals in the second.

Treat images as content to optimise, not just make compliant

High-resolution images, neutral background, product centred and clearly identifiable score better than technically compliant but ambiguous images. Source: support.google.com/merchants/answer/6324350.

Maintain a clean account history

Since Merchant Center artificial intelligence weighs account history in the review process, keeping the disapproval rate low and resolving policy issues promptly is a medium-term investment in faster, more favourable Merchant Center automatic reviews.

FAQ: Merchant Center Automatic Review and AI