Earlier this month a shopper opened a page buried in her Amazon account and found that Amazon’s AI had decided she “has flat buttocks.”
The evidence was a pair of shaping leggings.
Within days everyone was opening the same page. One person “has no friends.” Another “treats cats as biological children.” A third “can’t keep real plants alive.” The internet read it as a privacy joke. And they should. But we’d encourage brands to also read it as a ranking update.
The page is called About You. It launched quietly on May 13 alongside Alexa for Shopping and it holds everything Amazon has inferred about you from your purchases, searches, Lists, reviews and chats with the assistant. Amazon calls these “the personal details that shape your shopping recommendations.” If the results are personal, so is the ranking. It’s Amazon’s opinion of who you are.
Let’s say two people type “protein powder” at the same moment. One has years of running shoes, gym gear and almond milk in her history. The other bought a single tub as a gift in 2023. Amazon knows this, so it shows them two different pages. Alexa taught Amazon to read your listing and reason about it. About You gives it a memory of the customer too. An engine that can reason about a product and remember a person isn’t ranking anymore. It’s recommending.
Here’s a quick example. Let’s follow one customer. She buys trail running shoes from you. Amazon records the sale and writes a line in her profile: trail runner. That line stays for years. When she later searches for socks, a hydration vest, a headlamp, Amazon reads those searches through that line and the brand that created it has the inside track. If you only sell shoes, you created the line and someone else gets the socks. If you sell socks too, you get them. That’s why starter products and bundles are worth more than their margins say. They’re how your brand becomes part of who a customer is. And Amazon decides who your product is for by reading one thing, your listing. “Premium high-quality protein powder for men women adults” describes nobody. “Plant-based pea protein for lifters who skip dairy” describes a person Amazon can recognize.
Lumian’s Alexa Visibility tool shows you how Amazon’s assistant reads your products and where it’s left guessing.
For fifteen years sellers fought over Page 1. There is no Page 1 anymore. There’s one per shopper, written from a file you’ll never read. The only thing you control is the evidence.
Audit Your Entire Amazon Account in 15 Minutes with Lumian Agents.
Try Lumian Agents. [It’s Free]
In this week’s issue:
Marketplace: Walmart Agents, Review Requests, Holiday Returns
Seller Central Updates: Sponsored Prompts
Twitter/X Spotlight: About You
Marketplace Madness
What happened:
On Oct 21, Walmart CEO John Furner announced Agent Gateway, which lets approved AI agents, including ChatGPT, Gemini and Meta's Muse, search Walmart and Sam's Club inventory and buy on a shopper's behalf. Agent Profiles, with per-purchase approvals and spending limits, roll out in November. Amazon blocked Meta's Muse in September and still treats unapproved agents as a violation of its terms.
Why it matters:
An AI agent can now buy your Walmart listing but not your Amazon one. If your Walmart catalog is thin, stale or missing attributes, that is AI-driven demand you can't capture. Price, stock and product data on Walmart just became part of your AI visibility, not a side project.
What happened:
The Review Requests ad type announced at unBoxed is coming to US advertisers this fall, with an open beta starting in late October. It asks verified buyers to leave a rating, and it's only for products with fewer than 1,000 ratings; campaigns pause automatically at 1,000. Amazon says beta products got about 3x more ratings per week.
Why it matters:
For launches and stuck listings this could be the cheapest social proof on the platform. But the 3x figure comes from a short beta with no control group, so budget for it separately from PPC and measure cost per new review. Fix your top recurring complaint first; more reviews will surface it faster.
What happened:
Amazon posted on Oct 16 that anything bought between Nov 1 and Dec 31 can be returned until Jan 31, 2027 (Apple products by Jan 15) and that the window will repeat every year. It applies to FBA and FBM automatically, and sellers can't opt out. Peak fulfillment fees also started Oct 15, averaging about $0.32 more per unit through Jan 14.
Why it matters:
Your Q4 isn't finished until February. Don't set January ad budgets or reorder quantities off December revenue alone. Plan from sales net of expected returns, especially in gifting and size-sensitive categories.
Seller Central Update of the Week
Amazon is testing Sponsored Prompts in search.
The new ad placement in search results is a sponsored question that sits beneath the product listings.
Click it and it opens an Alexa for Shopping conversation, with answers and sponsored products mixed in, and shoppers can keep asking follow-ups without leaving search.
It’s a new way for brands to get in front of a shopper while they’re still deciding what to buy, and it blurs the line between an ad and a shopping assistant.
Twitter Spotlight
Meme Therapy
How did you like this week's newsletter?
Best,