A rare glimpse inside Google's performance-scoring system reveals the exact criteria the search giant uses to rank your listings. Here's what it means for your bottom line.
On September 7, 2026, Google's Merchant Center accidentally showed the prompts it feeds to its AI system when generating performance insights. Instead of displaying a clean summary of your top-performing products, the system spit out the raw instructions: 'thinking,' 'role,' 'formatting,' and other mechanical directives that Google's AI actually uses to rank and evaluate your catalog.
For e-commerce business owners, this is a rare window into the exact criteria Google applies when deciding which of your products get prominence in search results and shopping feeds. It's the closest thing to reading Google's own playbook on product performance.
Merchant Center has always been a black box. You upload product data, Google surfaces some products as 'top performers,' and you're left to guess what made them stand out. The insights reports were supposed to explain it, but they often felt vague or generic. Now, because of the prompt leak, the logic is exposed: Google's AI is checking data completeness, formatting consistency, and performance signals in a specific order and with specific weights.
This means sellers who understand and implement these criteria before Google patches the leak have a tactical advantage. Your product titles, descriptions, attributes, and feed structure can be optimized to align with exactly what Google's scoring system looks for, rather than hoping your data is 'good enough.'
The leaked prompts show that Google's AI evaluation includes role definition, formatting standards, and a 'thinking' phase where the system reasons through product quality signals. This suggests Google is not just running a simple checklist; it's using structured reasoning to compare your products against its own quality benchmarks.
The smartest move is to treat this leak as a temporary guide to Google's real standards. Pull a sample of your top-performing and underperforming products from Merchant Center, then reverse-engineer why Google ranked them that way using the prompt logic as your decoder.
Once you understand the pattern, you can rebuild your feed to match it. This isn't about gaming Google; it's about giving Google what it already wants to see, faster and more clearly.
Google will fix the prompt leak. The accidental exposure of internal AI instructions is too visible to leave alone. But the underlying evaluation criteria will remain the same, Google still cares about data quality, formatting, and performance. The difference now is that you've seen the thinking process behind it, and you can act on that knowledge before the window closes.
How WebKing runs this
We extract and apply Google's own performance criteria to your product data. Instead of guessing what Merchant Center wants, we align your feed structure and content to the exact prompts Google revealed, giving your best products the visibility they should have had from the start.
Google's system accidentally showed the behind-the-scenes instructions it gives its AI when grading product performance. These prompts include checks for data quality, formatting, and relevance that determine which products Google highlights as top performers in your Merchant Center account.
Your listings are already being scored by these same hidden rules. Now that the prompts are visible, you can align your product data, titles, and attributes to match what Google's AI actually looks for, potentially boosting visibility for your best sellers without waiting for Google to send you a generic insight.
Google will likely patch the prompt leak, but the underlying logic for evaluating product performance won't change overnight. The criteria around data completeness, formatting standards, and performance signals are structural to how Merchant Center works.
It's primarily a transparency breach that reveals Google's methodology. There's no indication of account data exposure or security compromise, just the accidental revelation of how Google's AI evaluates which products deserve top placement in search results.
Sources
The Lab is original analysis by WebKing. We summarize and interpret developments from the sources above for industrial, commercial, and small business owners. Figures are reported as published by their sources.
More from the desk
Photo: EqualStockA UK mattress brand tied customer messaging to purchase history and timing. The result: nearly three-fifths more customers ordering again in Q4 2024.
Photo: Zoshua ColahCore Web Vitals failures are killing leads for small businesses. We show you which design decisions tank rankings and how to fix them before your competitors don't.
Photo: Anastassia AnufrievaA major newsroom just reorganized its entire audience strategy because search is sending fewer clicks to their site. Here's what that means for any business relying on organic discovery.
Free audit
We read your search, ads and AI visibility, then tell you which of these moves actually applies to your business. Free, 30 minutes, no pitch deck.
