Google Shopping’s New AI-Curated Feed Is Forcing Merchants to Rethink Product Data Strategy

For the better part of a decade, winning on Google Shopping meant one thing: bid high, optimize titles, harvest search term reports. That playbook is breaking down fast. Google’s June 2026 rollout of its AI-curated Shopping carousels — now live across roughly 70 percent of U.S. product search queries, according to internal agency data reviewed by Ecommerce Times — is fundamentally restructuring how products surface, how impressions are allocated, and how DTC brands calculate return on ad spend.

The update, which Google has internally branded “Shopping Intelligence,” uses a multimodal ranking model that weighs feed quality signals — structured attributes, image resolution, review velocity, return policy clarity, and price competitiveness — alongside traditional bid data. For merchants whose Google Shopping strategy was built on aggressive CPCs and thin product data, the shift is brutal. For operators who invested in feed infrastructure, it’s a meaningful tailwind.

What exactly changed in Google’s June 2026 Shopping update?

The mechanics matter here. Google’s new carousel format, which appears above the traditional PLA grid on roughly 60 percent of high-intent commercial queries, pulls from a ranked pool of eligible products using what Google calls a “relevance quality score” — a composite metric that goes well beyond bid. According to documentation shared with Premier Partners in May, the score incorporates:

• Feed attribute completeness (all 77 optional attributes now carry ranking weight, not just the required seven)
• Image quality signals, including background removal, lifestyle context, and resolution above 800px
• Seller ratings aggregated from Google Customer Reviews, Trustpilot, and verified third-party sources
• Return and shipping policy structured data pulled directly from merchant center
• Price competitiveness index benchmarked against category median

The practical result: a merchant running a $12 CPC on a mattress topper with a complete, clean feed and 4.6-star seller rating can outplace a competitor bidding $22 on a thin feed with missing size attributes. Several agencies confirmed to Ecommerce Times they’ve seen CPC efficiency improve 18–34 percent for clients with strong feed infrastructure since the rollout, even as overall Shopping CPCs remain elevated.

“We had clients who were spending $80K a month on Shopping and suddenly saw impression share drop 40 percent in two weeks. The bids didn’t change. The feed did — or rather, it hadn’t changed in three years, and that was the problem.” — Caitlin Mercer, VP of Paid Search, Metric Theory

💡 Article Summary

Key Insights

1

What exactly changed in Google’s June 2026 Shopping update?

2

Which merchant types are getting hurt most by the new ranking model?

3

How are top-performing merchants adapting their feed and bidding strategy?

4

Is Performance Max still relevant alongside the new Shopping Intelligence carousel?

5

What should operators do in the next 30 days to protect Shopping performance?

Source: Ecommerce Times

Which merchant types are getting hurt most by the new ranking model?

The losers in this transition cluster around two profiles. First, mid-market brands that outsourced feed management to their agency years ago and never built internal product data infrastructure. Second, Amazon-first sellers who recently expanded to Google Shopping using catalog exports that were never optimized for Google’s attribute schema.

Andrew Faris, who runs AJF Growth and advises a portfolio of seven-figure DTC brands, flagged the issue in a widely-circulated LinkedIn post in late July. He noted that several brands in his network saw Google Shopping ROAS fall from 4.2x to 2.8x between June 1 and July 15 without any meaningful budget or bid changes.

“The algorithm is now effectively taxing lazy feed management. If your product titles are still in the format ‘Brand – Product Name – Color’ with nothing else, you’re leaving half your impression potential on the table. Google wants the full sentence.” — Andrew Faris, AJF Growth

Agencies using DataFeedWatch, Feedonomics, and GoDataFeed reported a surge in new client inquiries in July — Feedonomics alone said onboarding requests from Google Shopping-focused accounts were up 60 percent month-over-month, per a spokesperson. The common thread: brands that let feed optimization atrophy during years of Advantage+ Shopping success on Meta are now paying the price on Google.

How are top-performing merchants adapting their feed and bidding strategy?

The operational response from leading operators involves three parallel workstreams. First, a full feed audit against Google’s expanded 77-attribute schema, with particular attention to product_type taxonomy, item_group_id for variant clustering, and the newly-weighted shipping_label and return_policy_label fields. Second, image refresh programs — several brands are running structured shoots specifically to generate white-background and lifestyle variants that satisfy Google’s image quality signals. Third, aggressive seller rating cultivation, including Trustpilot integration and Google Customer Reviews enrollment.

Ronak Shah, co-founder of Obvi (a collagen supplement brand generating over $30M annually), told Ecommerce Times his team rebuilt their entire Merchant Center feed structure in June after seeing early signals from beta partner data. The rebuild took three weeks using Feedonomics and a dedicated data analyst, but yielded measurable results.

“We went from 68 percent feed attribute completeness to 94 percent in about three weeks. Our impression share on high-intent queries went up 22 percent. CPC actually dropped because we’re now eligible for the AI carousel placements that we weren’t touching before.” — Ronak Shah, Co-founder, Obvi

On the bidding side, the recommended approach from agencies is shifting toward tROAS strategies anchored to product-level margin data rather than blended account targets. With Google’s AI model already making placement decisions based on quality signals, layering smart bidding on top with clean margin inputs is producing stronger efficiency than manual CPCs for most account types.

Is Performance Max still relevant alongside the new Shopping Intelligence carousel?

This is the question consuming agency strategy calls right now. Performance Max campaigns still run across the full Google network and pull from Shopping inventory, but the AI carousel placements operate through a separate ranking layer that isn’t exclusively controlled by PMax budget allocation. Several agencies have started rebuilding Standard Shopping campaigns specifically to give Google’s AI carousel system cleaner, more attributable data signals — a counter-intuitive move given Google’s aggressive push to consolidate everything into PMax over the past three years.

Kirk Williams, founder of Zato Marketing and one of the most-followed Google Shopping practitioners in the agency community, published a detailed breakdown in late July arguing that Standard Shopping campaigns will have a meaningful role in 2026 account structures precisely because they give operators more control over the feed signals Google’s new ranking model is reading.

“PMax is a black box on top of a black box right now. The Shopping Intelligence layer adds a third layer of opacity. Running segmented Standard Shopping campaigns alongside PMax gives you cleaner data and more levers to pull when the AI carousel starts allocating your impressions in ways that don’t match your margin structure.” — Kirk Williams, Founder, Zato Marketing

The practical recommendation from Zato and several other agencies is a hybrid structure: PMax for brand and remarketing-heavy SKUs where audience signals are strong, Standard Shopping for high-margin, high-velocity SKUs where feed quality is the primary ranking lever.

What should operators do in the next 30 days to protect Shopping performance?

The operational checklist from leading practitioners is specific:

• Run a Merchant Center feed health audit — focus on completeness percentage and disapproval rate, target above 90 percent completeness on optional attributes
• Enroll in Google Customer Reviews immediately if not already active — seller rating below 3.5 stars is a confirmed negative ranking signal in the new model
• Audit product images for resolution (minimum 800x800px), background quality, and lifestyle variant availability
• Add structured return and shipping policy data to Merchant Center under the new Shipping & Returns tab, not just via policy pages
• Segment top-margin SKUs into dedicated Standard Shopping campaigns with tROAS targets set at product-level gross margin, not blended account ROAS
• Review price competitiveness against category benchmarks using Price Insights in Merchant Center — products flagged as “price high” are now algorithmically deprioritized in carousel placements

The broader implication for DTC brands and marketplace operators is that Google Shopping has quietly moved from a pure auction model toward a hybrid quality-auction system that more closely resembles Amazon’s A9 organic ranking logic than traditional paid search. Operators who built their Amazon success on rigorous catalog hygiene will find the translation relatively natural. Those who treated Google Shopping as a pure media buy are facing a steeper learning curve.

Feed management platforms are already responding. DataFeedWatch announced an expanded AI attribute suggestion engine in its July product update, specifically designed to fill optional Google attribute gaps using existing product description data. GoDataFeed similarly rolled out a “Google Readiness Score” dashboard in late July that maps merchant center completeness against the new 77-attribute schema.

The window to adapt is narrowing. With back-to-school spend already flowing and Q4 planning underway, operators who delay feed and bidding restructuring risk entering the highest-stakes quarter of the year with Google Shopping infrastructure that no longer matches how the platform allocates impressions. The merchants who treat this as an infrastructure problem — not a media buying problem — will have a structural advantage heading into November.