In partnership with
Amazon is letting sellers keep self-fulfillment, but with fewer excuses when the delivery promise breaks. The marketplace is not just selling demand. It is grading the operating system behind that demand.
In today's MarketMaze:
News
1️⃣ Amazon tightens seller shipping
2️⃣ Android’s default tax comes due
3️⃣ Agent payments need guardrails
Insights
4️⃣ Retail media’s AI shock
5️⃣ DTC needs a second engine
6️⃣ The funnel survives
LET’S ENTER THE MAZE!
1️⃣ News
The Maze: Amazon is tightening FBM, the seller-fulfilled shipping lane, by making delivery promises more measurable and less optional.
Germany's FBM sellers now face a 90% on-time delivery bar, with affected listings at risk of deactivation from September 1.
The UK update also targets handling-time buffers, with two-day defaults changing to one day from July 15 and Automated Handling Time coming into play from September.
The policy keeps self-fulfillment open, but it makes sellers operate closer to Amazon's own delivery standard or absorb the conversion and listing risk.
Why it matters: Amazon is not just selling traffic to merchants. It is setting the operating math behind that traffic.
🤝 From our sponsors
2️⃣ News
The Maze: Europe’s top court left Google’s EUR 4.1 billion Android fine standing, turning a long appeal into a final warning on mobile platform defaults.
The Court of Justice dismissed Google and Alphabet’s appeal, ending the case path for Android’s search, browser, and app-store bundling.
The original Commission case targeted how Google tied Search and Chrome to Play Store access and restricted alternative Android versions through device-maker agreements.
Google says it changed agreements after 2018, but Europe is still treating defaults as market power, not neutral product setup.
Why it matters: Mobile defaults steer search, apps, payments, and shopping journeys before retailers ever compete. Distribution is the margin layer.
🤝 From our sponsors
3️⃣ News
The Maze: Cross River and Stripe are turning agentic commerce into a payment-permission problem, not just a smarter shopping interface.
Cross River is working with Stripe on issuing infrastructure for AI-agent transactions, centered on single-use virtual cards.
Cross River's own card issuing page now points to virtual cards for AI agents, with bank sponsorship and advanced authorization.
Stripe Issuing already supports spending limits, merchant-category rules, geographic controls, and real-time authorization through webhooks.
Why it matters: AI agents can recommend products. Payment controls decide whether the agent is allowed to buy them without creating fraud, dispute, or liability chaos.
4️⃣ Insight
The Maze: Retail media leaders are no longer just worried about walled gardens. The bigger fear is AI search intercepting shoppers before retailers see the visit.
Zero-click/genAI search leads the Bain and EMARKETER survey at 36%, ahead of agentic ad-buying at 28% and RM fragmentation at 24%.
The same AI threat returns at 22% for path-to-purchase disintermediation, because discovery and checkout can increasingly happen inside AI-led search layers.
Only 28% of RMNs have incentives aligned with merchandising, so the industry is fighting a channel-control battle while its own retail and media goals still collide.
Why it matters: RMNs need incrementality proof and merchant alignment, not just more inventory. AI turns retail media from a traffic game into a visibility game.
5️⃣ Insight
The Maze: DTC winners often chased the same hot theme as everyone else. The difference was the operating engine built underneath the trend, not the mood board.
Warby Parker, Dollar Shave Club, Oura, Quince, and Rhode all matched the visible consumer theme of their year, but the winners added vertical integration, distribution, data, retention, or supply-chain compression.
The payoff was wildly uneven: Peloton is shown at a $44.4B peak, Oura at $11.0B, Quince at $10.1B, while 2025 has no clear AI commerce or TikTok Shop winner yet.
The live test is not whether a theme feels right; it is whether it improves repeat purchase, margin, distribution, or a defensible customer habit.
Why it matters: Trend-spotting gets a DTC brand noticed. The second engine decides whether the brand compounds or becomes another expensive lookalike.
6️⃣ Insight
The Maze: Social, AI, and livestream shopping look big until you put them beside US retail. EMARKETER's 2029 math says the funnel bends. It does not disappear.
EMARKETER's 2029 forecast puts US retail excluding auto and gas at `$5.89 trillion`, versus `$152.5 billion` for social commerce and `$144.5 billion` for AI platform-driven ecommerce.
Livestreaming ecommerce reaches only `$31.2 billion`, making the three "collapsed funnel" channels about 5.6% of the retail bucket even when combined.
Social commerce still needs entertainment to keep users engaged, while AI commerce needs unified inventory, pricing, and fulfillment data before it can act reliably.
Why it matters: Retailers should fund new demand surfaces, but not confuse influence with replacement. The winning system still spans feeds, search, stores, checkout, and returns.
🗞️ Quick hits
Everything else you should know
🛒 Grocery and quick commerce speed up
💳 Agentic payments leave the lab
Worldline, ING, and Visa completed a live agentic payment transaction in Europe, showing banks and card networks are testing how AI-mediated checkout gets authorized.
📣 Acquisition math gets uglier
🕷️ Discovery, fraud, and trust get messier
THAT’S IT FOR TODAY!
You’re the reason our team spends hundreds of hours every week researching and writing this email.
What do you think of this issue?
See you next time in the maze!
MarketMaze team
