Who lets the agent in

The Signal

Amazon began blocking Meta's Muse shopping agent on the night of September 20 after Meta declined to remove Amazon from the agent's scope; shoppers using Muse now see pop-ups saying its use violates Amazon's terms. Amazon's stated objections: Muse accessed the store without notice or authorization, did not identify itself as automated, and appeared to capture and store customer login credentials. The next day Shopify went the opposite way, enabling Shop Pay checkout inside Muse by default for roughly one million eligible U.S. merchants (opt-out through its Agentic sales channel), and its shares jumped. Days later at its Accelerate conference (September 22–24), Amazon announced sellers can manage their eBay, Shopify, TikTok Shop, and Walmart businesses from inside Seller Central at no extra cost — noting that more than 95% of its independent sellers already sell across multiple channels.

Why It Matters

The agentic-commerce map just got its first hard borders. Amazon is protecting the shopper relationship — the search results page, the sponsored placements, the first-party signal that powers the largest retail media business — by refusing unvetted agents, while simultaneously making itself the operating system for sellers everywhere else. Shopify, with no ad business on its storefronts to protect, is betting that being the default checkout for every agent is worth more than controlling the discovery layer.

The Opportunity

Brands now need a split agentic strategy: optimize product data and checkout for open ecosystems (Shopify, Muse, Google) while assuming Amazon's inventory, pricing, and conversion data stay walled off. Shopify merchants should review agent-order attribution, fraud rules, and promo logic now, because default-on means agent orders can arrive whether or not anyone made a decision. Multichannel sellers should weigh the convenience of running every channel through Seller Central against handing Amazon visibility into their Walmart and TikTok Shop businesses.

The Signal

Adobe's holiday forecast, released September 28, projects a record $275.1 billion in U.S. online spending from November 1 through December 31, up 6.7%, and expects AI-driven traffic to U.S. retail sites — measured by shoppers clicking through from AI tools — to rise 130% year over year. In an Adobe survey of 5,000 consumers, 77% of those who have used AI for online shopping said it made them more confident in the purchase and 69% said they were less likely to return the item. On September 16, Google's holiday update expanded its Merchant Center 'AI performance insights' — which benchmark a retailer's share of voice against competitors in AI Mode and AI Overviews — to businesses in the U.S., Canada, Australia, New Zealand, and India, and began rolling out new checkout capabilities for its Universal Commerce Protocol, including transferring an AI-built cart to the merchant's own site.

Why It Matters

For two years, AI-driven shopping has been a narrative without a scoreboard. Adobe's forecast rests on more than a trillion retail site visits, and Google is now giving merchants first-party measurement of visibility inside its AI answers — the two numbers every channel needs before it gets a budget line: how much traffic, and how much of it is yours. The lower-returns finding matters for margin too: if AI-assisted buyers keep what they buy, the channel is worth more than its traffic share suggests.

The Opportunity

Treat AI surfaces as their own channel this holiday, with their own owner, feed-quality standards, and reporting. Retailers with Merchant Center access should baseline AI share of voice now, before Black Friday, so there is something to compare against in December. And test Google's cart-transfer flow before peak traffic: it is the rare agentic feature that sends the shopper back to your own checkout — and your own customer data.

The Signal

On September 16 the Federal Reserve raised its benchmark rate a quarter point to 3.75%–4%, its first increase since July 2023, in a unanimous vote — and the median official now expects one more hike this year. The same morning, August retail sales came in up 1.2% to $773.9 billion, the strongest monthly gain since March. A week later the Mastercard Economics Institute forecast holiday retail sales (excluding autos and gas) up 5.5% from November 1 to December 24 — the strongest season in four years — with online up 11% and in-store up 3.6%, though coverage of the forecast noted inflation is doing a meaningful share of the lifting. And Challenger, Gray & Christmas projected retailers will add about 450,000 seasonal jobs this quarter, below last year's 461,500 — which was already the smallest seasonal gain since 2008.

Why It Matters

Retailers are hearing two contradictory signals and choosing caution: nominal spending is strong enough to justify a record forecast, but a tightening Fed raises the cost of every credit-card-funded basket right as the season starts. The hiring number is the most honest of the week's data, because it is a commitment rather than a projection — and retailers are committing to fewer seasonal workers for a third straight year.

The Opportunity

Plan the holiday to the gap between dollars and units: price-driven topline growth means promotions, financing offers, and value messaging will do more work than full-price brand plays. Leaner store and fulfillment staffing raises execution risk on pickup and delivery during the traffic peaks retail media campaigns create — coordinate campaign timing with the retailer's labor calendar. And watch for the second hike the Fed penciled in: a December move would land squarely on the final shopping weeks.

The Signal

General Mills reported fiscal first-quarter results September 23: net sales of $4.4 billion, down 3% on its U.S. yogurt divestiture, with organic sales flat; adjusted EPS of $0.75 beat the $0.72 consensus, and the company reaffirmed a full-year outlook of organic sales between -1.5% and +0.5%, with input-cost inflation running at the high end of its 4%–5% range. CEO Jeff Harmening credited 'stronger product innovation and renovation focused on the benefits consumers are looking for today.' A day later Costco reported fiscal fourth-quarter sales of $93.9 billion, up 11.2%, with EPS of $6.75 — and disclosed $184 million in tariff refunds (including $10 million of interest), part of which it reinvested in lower prices on everyday Kirkland Signature items, which it says already run 15%–20% below comparable national brands. Costco's digitally enabled sales grew 19.5%.

Why It Matters

Two weeks ago Campbell's and Conagra bet on list-price increases. General Mills' print is the first test of the alternative: hold price, invest in product and brand, and accept flat organic sales for now. Meanwhile the biggest club retailer is converting tariff refunds into lower private-label prices — widening the gap national brands must justify at exactly the moment they are asking retailers to accept increases.

The Opportunity

Brands taking price should expect retailers to cite refund flows and private-label price cuts in every cost negotiation this fall — prepare the value story, not just the cost story. The General Mills path of renovation over pricing needs media to work: new benefits have to be discovered at the digital shelf, which makes retail media the delivery mechanism for innovation rather than a promotional afterthought. And Costco's 19.5% digital growth makes its still-young ad business worth a closer look in 2027 planning.

The Signal

On September 16, the court unsealed Judge Leonie Brinkema's full 106-page remedies opinion in the Justice Department's ad-tech case, two weeks after she rejected a breakup. Google keeps its AdX exchange, but must build integrations connecting AdX and its DFP publisher ad server to the open-source Prebid system, submit AdX bids to rival publisher ad servers on the same terms DFP gets them, share real-time win-and-loss bid data with publishers, and stop AdWords from bidding preferentially into its own tools. The remedies apply worldwide and run six years under a court-appointed technical monitor with access to Google's staff, systems, and source code — the DOJ had asked for 15. Coverage of the order puts the Prebid build at roughly 12 to 15 months.

Why It Matters

This is the first time publishers, sell-side platforms, and advertisers can plan against the actual compliance text rather than headlines. Behavioral remedies with a long build timeline mean the programmatic auction won't change materially before late 2027 — so the open web that retail media networks rely on for offsite reach stays largely on Google's rails through the next planning cycle, even as the largest independent DSP struggles.

The Opportunity

Publishers and SSP partners should start ad-server and header-bidding evaluations now, so they can act the moment the Prebid integrations ship. Retail media networks building offsite programs should note that the equal-terms and bid-data provisions eventually make it easier to route commerce-audience demand to publishers outside Google's stack. For 2027 plans, assume the status quo — but build optionality for 2028.