STUMP BEZOS
What percentage of all creator influenced purchases happen on Amazon?
[ Answer at bottom of email ]
👀 AMAZON’S SEARCH BAR JUST LEARNED TO TALK
For almost 30 years, Amazon trained all of us to talk like robots. We learned to type "mens running shoes size 11 black" instead of what we actually mean: "I need comfortable shoes for my dad who walks 3 miles a day and hates flashy colors."
Now the robot has learned to understand us. Andrew Bell recently went deep on how Alexa for Shopping sits behind the same old search bar, but when you type a whole messy human thought, the AI does the hard work. It runs multiple searches for you, compares products, double-checks facts, remembers what you like, and explains why it picked what it picked.
Two top Amazon execs (Doug Herrington and Rajiv Mehta) just laid out the roadmap. Herrington's money quote: in five years:
"The house is kind of reordering its regular supplies for itself and we're going to think that's totally normal."
Your fridge ordering its own filter isn't sci-fi anymore, it's the last step of a machine Amazon is building right now, inside the search bar you use every day.
The big shift in one sentence: The shopper's job shrinks to stating the need. Amazon's job grows to searching, comparing, verifying, and choosing.
Why this matters for your listings
When one shopper request quietly becomes five different AI-generated searches behind the scenes, your product either shows up across those paths or it doesn't exist. And two people typing identical words now get different results, because Alexa reshapes the candidate pool with remembered preferences before ranking even starts.
Actionable Nuggets
SEO isn't dead — it multiplied. Alexa decomposes one request into multiple search routes (form, material, style, use case, price). Your listing needs linguistic coverage across ALL the ways a need can be phrased, not just your top 10 keywords.
Make your PDP "answerable," not just pretty. Alexa's #1 use case is answering basic questions: Is it waterproof? How wide is it? Does it include batteries? If the answer is buried in one image or deep in A+ content, the AI can't confidently repeat it, and an unanswered question kills the sale.
Kill your contradictions. A dimension that says one thing in the bullets and another in the attributes creates a fact the AI won't trust or state. Audit for consistency across title, bullets, attributes, A+, and images.
Fill every structured attribute field. The AI leans on machine-readable data to verify claims before recommending you. Empty backend fields = invisible product.
Your reviews are now fact-checkers. Alexa cross-references your claims against review sentiment. If reviews contradict your "durable" claim, that claim becomes unsafe for the AI to repeat.
Don't overclaim. Alexa will literally tell shoppers "this isn't an exact match" and offer alternatives. The winning move is being safe to recommend — accurate, specific, defensible facts beat keyword-stuffed fluff.
Recommendations come with receipts. Alexa explains why it picked a product. Products with clear, verifiable differentiators get explainable recommendations; vague ones don't get picked at all.
Bottom line: Stop optimizing only for a ranking on a results grid. Start optimizing for an AI salesperson who reads your product data out loud to a customer, and only recommends what it can verify.
Turn Every Order Into More Reviews
Selling on Amazon, TikTok Shop, Walmart, or multiple marketplaces?
GetReviews helps you collect, manage, and syndicate your customer reviews across the channels where your products are sold.
And now, with our new TikTok Shop integration, your reviews can be automatically syndicated to TikTok Shop.
Don't wait - improve your review strategy today.
Use code 1FREEONME to get your 1st month free
🔭 YOU GOTTA SEE THIS
Despite the massive growth of e-commerce, 83% of all US retail spending still happens in physical brick-and-mortar stores.
In this episode of Marketing Misfits, hosts Norm Farrar and Kevin King sit down with Korie Minkus, the founder of Rock Your Product, to discuss how successful Amazon and Shopify brands can confidently make the leap onto physical retail shelves.
Korie breaks down her proven 5-step Retail Operating System, explaining exactly what buyers at major retailers are looking for, why your current packaging will probably fail in stores, and how to solve the massive cash flow challenges that come with fulfilling large purchase orders.
Whether you want to land your product in Target, Costco, or independent local boutiques, this episode is a masterclass in CPG retail distribution.
🔥 Join the Marketing Misfits Newsletter
🌎 INTERESTING STATS
🔒 AMAZON is LOCKING its DATA in the VAULT
Imagine your kid has the best trading card collection in school, and suddenly everyone (Google, AI companies, researchers, scrapers) wants to photograph his/her cards so they can build their own collection without doing the work. What does your kid do? She stops showing them off. That's exactly what Amazon is doing with its shopping data right now.
Here are the two big changes shoppers are noticing:
1. Order emails got weirdly vague. Since July, Amazon's confirmation emails no longer tell you what you bought. Instead of "Your CeraVe Moisturizer has shipped," you get "Your Beauty item is confirmed!" with a generic clip art picture. Want to know what's actually coming? You have to open the Amazon app and check "Your Orders."
Why? Because AI tools like Google's Gemini can read your inbox. Google is building AI shopping agents that mine your emails for purchase details, then use that data to recommend products and eventually shop for you across other stores.
Every detailed Amazon receipt sitting in a Gmail inbox is free intelligence for Amazon's biggest competitor. Vague emails starve Google of that data and force you back into Amazon's app, where Amazon controls everything. Amazon's official line is that it's about "customer privacy," but the timing tells the real story.
2. Reviews are being rationed. For months, shoppers have reported only being able to see a handful of reviews (some as few as 8) on a listing.
Click "request to see more reviews" and Amazon says you'll get an email within 5 business days.
Amazon admitted to Fast Company that this is about stopping "unauthorized data scraping." Its systems sometimes flag real humans as AI bots, and those humans get locked out of the review section.
The bigger picture: Tech companies call this "the DoorDash problem." If AI agents shop for people, nobody visits your website anymore, and whoever owns the customer relationship owns everything.
Walmart, Target, and Wayfair decided to join Google's AI shopping features. Amazon went the opposite direction. It sued Perplexity for letting AI shop on Amazon (and lost earlier this month), built its own AI assistant (Alexa for Shopping) so shoppers never need ChatGPT or Gemini, and is now walling off the data that makes its store valuable: what people buy and what people say about products.
What it means for sellers:
Review visibility is now unreliable. A shopper who can only see 8 reviews might see your one bad review with none of the good ones drowning it out. One misunderstanding review just got more dangerous.
Some buyers are walking. Shoppers say they won't buy what they can't research, and some are heading to other sites. Watch your conversion rates.
Your listing content matters more than ever. If shoppers can't lean on reviews, your images, bullets, and A+ content have to answer every question and objection on their own.
Post-purchase emails and actions you send through Amazon just got more valuable. Amazon's own emails no longer remind buyers what they ordered, so branded inserts, packaging, and follow-ups are doing more of the memory work.
The walls are going up everywhere. Expect scraping-based tools (review analyzers, competitor trackers) to get less reliable as Amazon tightens the gates.
Amazon spent 30 years collecting the world's best shopping data. Now that data is the prize in the AI wars, and Amazon is done sharing it for free, even if that means annoying its own customers a little along the way.
🛠️ BDSN FEATURE of the DAY 🛠️
Free Supply Chain Audits - 25 only
One dinner with his factory saved an Amazon seller $400K a year. Another got 60 day payment terms after one factory visit, unlocking $3M in sales over 12 months.
The China Magic team is behind both, and they just opened 25 free supply chain audits. Terms, pricing, margins, returns, MOQs and quality, benchmarked in one 30 minute call.
25 slots only. Claim your free audit here.
🤖 50% of SITES CAN’T BE BOUGHT FROM BY AI AGENTS
Researchers at Acom scanned 235 live product pages across 116 stores (Shopify, WooCommerce, Magento, BigCommerce) and asked one question: could an AI shopping agent actually find this product, understand it, trust the store, and buy it?
The median page scored 66 out of 100, which sounds fine until you see what's under the hood. 47% of pages are missing every single thing an agent needs to complete a purchase. As the study puts it: stores are readable, but almost none are buyable.
Why the scores are misleading: Most pages cluster in the 60s and 70s because everyone passes the same basic checks (page loads, AI crawlers allowed in, price visible) and everyone fails the same advanced ones. A page can score a healthy-looking 70 and still be a page no AI agent can transact with.
The kicker: Most of this information already exists on the page. It's written for humans in the product description or footer. It's just never published as structured data a machine can read. Four out of five major AI engines don't run JavaScript, so anything loaded after the initial HTML is invisible to them.
Actionable nuggets for your DTC site
Check for silent blockers first. 8% of pages had a leftover "noindex" tag (usually from staging setups or plugin defaults) telling search engines and AI to ignore them. Also check your robots.txt isn't blocking GPTBot, ClaudeBot, PerplexityBot, or OAI-SearchBot. These are one-minute fixes.
Make sure name, price, and availability are in the raw HTML, not injected later by JavaScript. If the agent can't see it on first load, it doesn't exist.
Grab the two cheapest wins: shipping and returns. You already wrote them for humans. Publish them as structured data (schema markup). Since 92 to 94% of your competitors haven't, this is the fastest way to jump the field.
Add real barcodes (GTIN/UPC), not just your internal SKU. Agents need them to match your product across the web and compare prices.
Declare your product category and specs as data. Nobody in the entire study did this. It's a wide-open lane.
AI agents are coming to shop, and right now they're showing up to stores where the doors look open but the register doesn't work. The gap between the middle of the pack and the top is an afternoon of fixes, not a rebuild. Do them before your competitors read this same study.
🔥 MORE HOT PICKS 🔥
🥃 PARTING SHOT
“An entrepreneur tends to bite off a little more than he can chew hoping he’ll quickly learn how to chew it.”
✌🏼 Have a great weekend.
See you again on Monday.
The answer to today’s STUMP BEZOS is
94% of all creator influenced purchases are on Amazon
