Product-Level Data Quality

Google AI Max for Shopping Reads Your Feed, Not Your Bids

by

Google launched AI Max for Shopping on April 30, 2026 as a one-click upgrade for any standard Shopping campaign. The system reads Merchant Center feed attributes — Google specifically named fabric softness, material durability, and fit — to generate ad copy, pick landing pages, and select ad formats. The default Google for WooCommerce plugin emits a barebones feed (title, price, image, GTIN). An independent 250-campaign study found AI Max delivered approximately 35% lower ROAS when feed and conversion-data quality were thin. WooCommerce stores need 30 days of feed enrichment before flipping the upgrade.

AI Agents Need a Stable SKU. Your WooCommerce Variants Don’t Have One.

by

WooCommerce 10.3 enabled the Model Context Protocol on October 3, 2025, and Google’s Universal Commerce Protocol now demands a stable SKU on every sellable variation that never changes once published. Most WooCommerce variable products do not meet that bar — they share parent SKUs across variations, drift attribute taxonomies, use free-text categories, and rely on inventory sync that exceeds UCP’s sub-50ms latency budget. Human shoppers tolerated that ambiguity. AI agents will not. The failure mode is silent: variants disappear from agent results, agent transactions abort mid-checkout, and store owners only find out via the missing revenue line. The fix is a four-step variant audit before agentic traffic arrives, not after.

The One WooCommerce Fix That Tells You What to Stock Next

by

The single most valuable data-quality fix for most WooCommerce stores is ensuring every purchase event captures the full product variant — size, colour, material, bundle — not just the parent product SKU. Roughly 60% of WooCommerce stores don’t capture this, which blinds them to variant-level demand at exactly the granularity that inventory decisions are made. Once the variant is captured on every purchase event, a store owner can ask their data (or Claude connected to BigQuery) which variants are selling fastest, which are sitting dead, and which parent products have one hot variant hiding five cold ones. Variant-level data is roughly 3x more useful for stocking decisions than parent-product data.

Your WooCommerce items Array Is Sending Google Product Data It Cannot Use

by

Your GA4 purchase events are firing. Revenue numbers look reasonable. You’ve checked the thank-you page trigger, verified the transaction ID, compared the total value. Everything appears correct. But 70% of ecommerce stores have incomplete tracking configurations — and the most invisible failure point isn’t the event itself. It’s the product data inside it. Every GA4 … Read more