In partnership with
Happy Thursday! Coupang rejected four South Korean antitrust inspections, then sued. The fight tests how regulators can investigate who really funds a platform's lowest price.
News
1️⃣ Coupang locks out the regulator
2️⃣ eBay closes the airbag aisle
3️⃣ Anthropic books tomorrow's power
Insights
4️⃣ Attribution flips channel ROI
5️⃣ Commerce media escapes retail
6️⃣ eGrocery growth moves inward
LET’S ENTER THE MAZE!
1️⃣ News
The Maze: Coupang rejected four South Korean antitrust inspections into claims that its automatic price-matching coupons shifted costs to suppliers, then sued.
Coupang sued to stop the probe, arguing that regulators failed to provide seven days' notice.
KFTC says specialized retail-law powers or the evidence-destruction exception allow surprise inspections.
Suppliers allege that automatic price-matching coupons cut settlements while Coupang disputes any change to supplier payments or taxes.
Why it matters: The court may shape how regulators inspect powerful platforms, while the underlying question—who funds the lowest price—remains open.
2️⃣ News
The Maze: eBay will ban standalone airbags and inflators from September 24. Its certified-seller system is becoming a global category-wide safety stop.
The policy covers every seller, while non-inflator parts remain allowed under different U.S. and international seller rules.
NHTSA links defective `DTN60DB` replacements to ten deaths and three severe injuries, with normal VIN recall searches unable to identify the aftermarket parts.
eBay previously had 218 approved sellers using certification, donor-VIN, recall, condition, and hazardous-shipping controls; legitimate recyclers now lose the channel too.
Why it matters: When a marketplace cannot trace a safety-critical part to its final user, removing the category can cost less than policing every listing.
3️⃣ News
The Maze: Anthropic reportedly committed $45 billion over six years for 460 MW of future Nscale compute—a $7.5 billion yearly average before payment timing.
Nscale would supply an integrated West Virginia stack: onsite power, cooling, networking and Nvidia Vera Rubin systems, not just loose GPUs.
Reported capacity starts in late 2027, while Nscale's current phase-one target is early 2028; the private contract leaves milestones undisclosed.
Anthropic already secured up to 5 GW with Amazon and multiple Google TPUs, turning supplier diversity into a portfolio of giant obligations.
Why it matters: More compute can make Claude services more reliable. It also leaves Anthropic needing enough profitable usage to fill capacity contracted years in advance.
4️⃣ Insight
The Maze: One source case gives Google Ads £8.20 per £1 in platform attribution but £3.40 in MMM. Social moves the other way. The budget ranking flips.
Google falls by £4.80 per £1 between the two measurement views, leaving platform attribution at 2.41 times its MMM estimate.
Meta rises from £2.80 to £4.20, while TikTok moves from £1.90 to £2.80 and Snap from £1.80 to £2.60.
A stronger stack combines MMM with incrementality tests instead of treating the final observable interaction as the full cause.
Why it matters: Attribution helps run campaigns. MMM and experiments test contribution. When their rankings diverge, the budget needs investigation—not autopilot.
5️⃣ Insight
The Maze: Nonretail networks rise from 13.0% to 17.4% of US commerce-media spending by 2030. Retail keeps scale, but no longer owns every useful buying signal.
Retail remains dominant at 82.6% in 2030, yet nonretail gains 4.4 points inside a market forecast to reach $142.07B.
Delivery, finance and travel networks can monetize movement, payment and intent data without owning the final checkout, but each needs a distinct playbook.
Advertisers already use a median of six commerce-media networks, while just 3% say audience incrementality is measured very accurately in McKinsey's survey.
Why it matters: First-party data opens the door, not the budget. Networks must turn unique signals into simple buying, comparable metrics and incremental sales.
6️⃣ Insight
The Maze: June 2026 U.S. eGrocery growth came from active shoppers ordering more often and spending more, while total user growth nearly stopped against a strong base.
Order-frequency growth accelerated from roughly 3.6% in 2024 to 9.9% in 2026, extending a 22-month year-over-year streak.
Average-order-value growth rose from about 0.5% to 8.1%, while Amazon widened Same-Day baskets to fresh food and everyday goods.
Total active-user growth fell to around 0.2%, so MAU now hides the real contest: frequency, basket, primary-store status, and fulfillment cost.
Why it matters: Online grocery has become a share-of-wallet fight. Retailers need more missions and better economics from existing shoppers, not just another MAU headline.
🗞️ Quick hits
Everything else you should know
⚖️ Platforms meet new guardrails
Meta agreed to pay up to $18 billion and redesign teen experiences across its apps, setting an expensive product-governance benchmark for attention platforms.
Shein lost a copyright case against Temu over copied fast-fashion designs, exposing the difficulty of defending supplier-generated assortment while competing on speed and similarity.
💸 Merchants reprice the transaction
eBay will default selected calculated-shipping listings to discounted label rates from October 1, lowering checkout prices unless affected sellers opt out by September 25 and keep the carrier savings.
Washington University research found buy now, pay later can influence retailer prices and inventory decisions, making the payment method a merchandising input rather than only a conversion tool.
Curology says Walmart shoppers convert to prescription skincare at five times its online rate, turning a $24.97 shelf product into a stronger telehealth funnel.
🌍 Reach expands through partners
Marks & Spencer expanded through Zalando to 22 European markets, gaining cross-border reach without building a separate local ecommerce stack in every country.
Advance Auto Parts rebid carrier contracts and expects tens of millions in transportation savings, turning procurement and distribution discipline into a material margin lever.
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
