AI surfaces are becoming addressable ad inventory at roughly the same pace classical SEO surfaces are shrinking. OpenAI shipped self-serve Ads Manager Monday and went live with paid Product Feed Campaigns days later. Google killed FAQ rich results in between. Etsy went native ChatGPT app. Visa and Mastercard stood up agent payment rails in Canada and Portugal on the same day. ICSC and McKinsey put $1T on US agentic commerce by 2030; Sopra Steria put €310B on Europe within a decade.
TL;DR
- ·OpenAI ships self-serve ChatGPT Ads Manager (CPC, pixel, Conversions API, $50K floor dropped) on May 5, and goes live with paid Product Feed Campaigns: datafeed upload via the merchant program, Feed ID selection, attribute-based filtering (category, brand, seller, rating, color, condition), and dynamic creative generated from product attributes.
- ·Etsy launches as a native ChatGPT app, exposing 100M+ listings inside conversation via @Etsy.
- ·Google drops FAQ rich results from Search on May 7. The FAQPage schema itself is not deprecated and still feeds ChatGPT, Perplexity, Gemini, and AI Overviews, so keep the markup.
- ·Visa Agentic Ready expands to Canada with BMO, CIBC, RBC, Scotiabank, and TD signed up.
- ·Mastercard opens a 600-person Lisbon Centre of Excellence and runs Portugal's first live Agent Pay transaction.
- ·ICSC + McKinsey: US agentic commerce projected to hit $1T by 2030; 68% of US consumers already used an AI tool to shop in the past three months.
- ·Sopra Steria: up to €310B of European e-commerce will flow through AI-assisted journeys within 10 years, but 41% of Europeans don't trust any organization to provide an AI shopping agent.
OpenAI ships self-serve ChatGPT Ads Manager and starts testing Product Feed Campaigns
On May 5, OpenAI opened beta access to a self-serve ChatGPT Ads Manager for US advertisers. CPC bidding joins the existing CPM model, a pixel and Conversions API ship for tracking, and the $50K test-spend minimum is gone. Expansion preview: UK, Mexico, Brazil, Japan, South Korea. Ecosystem partners include Pacvue, Kargo, StackAdapt, Criteo, Adobe, plus Dentsu, Omnicom, Publicis, and WPP.
Days later, paid Product Feed Campaigns showed up live in the same Ads Manager, first surfaced by Marketplace Pulse founder Juozas Kaziukėnas in a LinkedIn post titled "Coming next to OpenAI Ads: Product Feed Campaigns". Advertisers connect a product datafeed via OpenAI's merchant onboarding (running against the published Product Feed Spec), select the feed by ID, and either run ads against the entire catalog or carve out a subset using product-attribute filters (category, brand, seller, rating, color, condition). Creative is generated dynamically by merging product attributes with templates, the same Performance Max and Google Shopping mechanic ecommerce teams already understand. The combined release closes the loop: addressable ad surface plus merchant-managed catalog plus dynamic creative, all live inside ChatGPT, on roughly a ten-year compressed timeline versus Google's Shopping plus Merchant Center build-out.
Etsy launches a native ChatGPT app, opening 100M+ listings to conversation
Etsy went live in beta as a native ChatGPT app on May 5. Users tag @Etsy in a prompt and see listings rendered inside the conversation. Second meaningful ChatGPT bet for Etsy after the Instant Checkout pilot ended in March.
Native apps differ from catalog feeds in a critical way. Feeds let an LLM mention your products in text. Apps let your brand render its own UI inside the conversation, with prices, photos, and click-through paths you control. More on what the app surface unlocks in our personalize-answers-in-ChatGPT breakdown.
Google formally drops FAQ rich results, but the schema markup still feeds AI search
On May 7, Google updated its Search Central documentation to mark FAQ rich results as no longer supported. Phased rollout: the rich-results report and Rich Results test lose support in June, Search Console API drops FAQ data in August. Second tranche after January, when seven other types (Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement, Vehicle Listing, Practice Problems) lost SERP rendering.
The FAQPage schema itself is not deprecated. Google still parses it, and ChatGPT, Perplexity, Gemini, and AI Overviews all consume the same markup to ground answers. Stripping it costs you AI citations for zero upside. Keep the schema, drop FAQ blocks you only kept for the rich result, rewrite answers as something a model would actually quote. Classical rich results contract; AI surfaces expand. Budget should follow.
Visa Agentic Ready expands to Canada with all five major banks signed up
On May 5, Visa brought Agentic Ready to Canada with BMO, CIBC, RBC, Scotiabank, and TD signed up to test agent-initiated card flows in production. Visa frames it as a step toward 100% online-payment tokenization. One week after the 10-market APAC rollout. Visa is the only card network publicly running all three plays in parallel: FIDO standards work, issuer enablement, and merchant-side integration.
Mastercard opens a 600-person Lisbon hub and runs Portugal's first live Agent Pay transaction
Mastercard opened a 600-person Centre of Excellence in Lisbon on May 5 and used the event to execute the first live Agent Pay transaction in Portugal with a Portuguese-issued card. The hub focuses on tokenization, agentic commerce, and stablecoins. Visa moved on Canada the same day. Two networks, two new countries, one week.
ICSC and McKinsey: US agentic commerce projected to hit $1T by 2030
ICSC and McKinsey's May 5 report forecasts US agentic commerce at $1 trillion by 2030. 68% of US consumers have used at least one AI tool to shop in the past three months; 62% used AI to compare brands, models, prices, or reviews. The $1T figure is now the canonical 2030 US TAM number every pitch deck will cite for the rest of the year. The 68% adoption rate also runs well ahead of the European number Sopra Steria reported the same week, consistent with the "US adopts faster, Europe regulates faster" pattern visible across consumer AI.
Sopra Steria: €310B European agentic commerce TAM, but 41% don't trust AI shopping agents
A Sopra Steria study of 8,400 consumers across eight European countries projects up to €310 billion of UK and European e-commerce flowing through AI-assisted journeys within a decade. 41% of Europeans don't trust any organization to provide an AI shopping agent. The trust gap is the part most TAM forecasts skip. It explains why payment-network-led approaches (Visa Agentic Ready, bank-affiliated agents) are out-pacing direct-to-consumer AI shopping platforms in the EU. Trust is being borrowed from existing institutions, not built from scratch.
