In partnership with
Happy Friday! Amazon is putting product titles on a 75-character diet and letting AI write the replacement when sellers miss the limit. Brands get 14 days to intervene, turning catalog copy into a timed operating process.
News
1️⃣ Amazon puts titles on timer
2️⃣ Mastercard guards the agent checkout
3️⃣ OpenAI widens AI’s cheap lane
Insights
4️⃣ One average misreads conversion
5️⃣ Amazon’s store became the front door
6️⃣ Retailers miss the app moment
LET’S ENTER THE MAZE!
1️⃣ News
The Maze: Amazon now caps most product titles at 75 characters. Its AI can supply the replacement, while brand owners get 14 days to intervene.
Amazon's new structure keeps 75 characters for the title and adds 125 searchable characters for Item Highlights.
Over-length titles move gradually to AI recommendations, with brand owners able to review, modify or approve them in Review Listing Changes before publication.
Sellers keep responsibility for accuracy, while Amazon controls the default copy and deadline—a risky trade when one title must carry brand, variant and search intent.
Why it matters: Amazon is cleaning up its mobile shelf by automating catalog copy. Brands now need a review operation, not just a title-writing rule.
2️⃣ News
The Maze: Mastercard argues that cards can survive AI shopping by owning the identity, permission and risk layer behind every agent purchase.
Agent Pay gives software a tokenized credential and an auditable record of what the user allowed it to buy.
Agent Pay for Machines adds spending rules and multi-rail settlement for continuous, low-value software transactions.
A live European payment proved the plumbing, but still required explicit consumer approval.
Why it matters: AI platforms may own the shopping interface. Mastercard wants banks and card networks to keep the more valuable checkpoint: trust, authorization and liability.
3️⃣ News
The Maze: OpenAI cut GPT-5.6 Luna prices by 80% and Terra by 20%, making high-volume agent work far cheaper only three weeks after launch.
Luna now costs $0.20 per million input tokens and $1.20 per million output tokens, while Terra costs $2 and $12.
OpenAI links the cuts to better routing, GPU kernels, speculative decoding and caching; its kernel work reduced serving costs by 20%.
Commerce teams can route repetitive catalog, support and workflow steps to Luna, but long prompts, retries, tools and quality checks still determine the real bill.
Why it matters: Cheaper models expand viable automation, but the advantage goes to teams that measure cost per successful task and route work by risk.
4️⃣ Insight
The Maze: Global ecommerce converts at 2.68%, but Switzerland reaches 4.14%. The useful target starts with market and category, not one tidy average alone.
Switzerland leads the top five by 1.46 points over the global benchmark; even fifth-place France reaches 3.36%.
Grocery converts at 4.39% versus 1.94% for Furniture & Homeware, a 2.45-point gap created by purchase frequency, price and deliberation.
ECDB’s six-market report mean is 3.38%, showing how the country set alone can rewrite the average.
Why it matters: A bad benchmark creates fake funnel problems. Set targets by market, category and traffic context before spending on discounts or redesigns.
5️⃣ Insight
The Maze: Amazon's online stores produced 37.6% of 2025 revenue. The other 62.4% came from seller fees, AWS, ads, subscriptions, physical stores, and other lines.
Online stores led at $269.3 billion, followed by seller services at $172.2 billion and AWS at $128.7 billion.
Advertising grew 22.1% and AWS 19.7% in 2025, while online stores grew 9.0% in Amazon's full-year results.
Third-party services are Amazon's net commissions, fulfillment, and shipping fees—not the full merchandise value sold by marketplace merchants.
Why it matters: Amazon does not need to own every product to monetize demand. It can earn from supply, discovery, loyalty, fulfillment, and infrastructure.
6️⃣ Insight
The Maze: App-based shopping is mainstream, but owned retail apps are not: DHL finds 72% of shoppers are app-first versus 38% of businesses with an app.
DHL surveyed 29,000 shoppers and 5,800 businesses, revealing a 34-point gap between app preference and proprietary app availability.
Businesses sell across three platforms on average, yet their owned-app presence trails websites at 65%, social media at 63%, and marketplaces at 54%.
An app is only useful when the operating model works: 62% of shoppers abandon purchases when their preferred payment option is missing.
Why it matters: Mobile strategy is not “build an app.” It is owning repeat customer utility where frequency pays, while removing checkout and fulfilment friction everywhere.
🗞️ Quick hits
Everything else you should know
🧑💻 AI assistants move onto the shop floor
Japanese retailer Yamada Denki deployed an OpenAI-powered multilingual retail agent that operates around the clock, showing real-time AI assistance moving from demo to scaled store operations.
Kohl’s launched an AI shopping assistant for product discovery and purchase decisions, testing whether conversation can lift retail conversion.
🛍️ Social commerce reaches seller scale
TikTok Shop says 300,000 UK small businesses now sell on the platform as live shopping grew 55% year over year, making social commerce a scaled seller channel.
🏗️ Automation absorbs more capital and volume
Nexus Data Centers is reportedly seeking $15 billion for a Google-backed Anthropic computing facility, exposing the capital intensity behind AI-mediated commerce.
UPS says automated facilities now handle more than two-thirds of its United States package volume, changing parcel costs and network flexibility for ecommerce shippers.
🎯 Discovery gets closer to checkout
Google added holiday Demand Gen upgrades including target-return-on-ad-spend bidding and direct checkout links, tightening the path from product discovery to conversion.
💳 Cross-border payments speed up
Swift is building a real-time cross-border payment corridor with Canadian financial institutions, aiming to reduce settlement friction for merchants, suppliers and international buyers.
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
