COHORT STARTS THIS WEDNESDAY - DON’T MISS IT!

Most AI product images still look like AI product images. Wrong labels. Hallucinated packaging. Sizes Amazon quietly suppresses.

We wrapped a session last week with John Aspinall (Velocity Sellers / Aspie AI), and he gave away his AI system —> Amazon-ready hero images, full 6-image stacks, Premium A+ modules, social ad packs, and video. All of it inside the ChatGPT Codex and Grok.

A few things he showed (all in Youtube replay above):

👉 Watch the replay above

If you want to actually build this with John in four live 2-hour Wednesday sessions, real ASIN, finished assets every week, plus the skills themselves to keep, click here

Doug Herrington, Amazon's CEO of Worldwide Stores, keeps a 1910 Sears and Roebuck catalog in his office. Roughly 20,000 items, everything from socks to entire houses, each with a title, a blurb, and a number you wrote on an order form in the back.

His point in bringing it to meetings for years: every retail format has a ceiling. Sears could not scale to 200,000 items because you cannot mail someone a book that size and expect them to shop it.

On his Learn and Be Curious podcast, Herrington argued that Amazon's own format has now hit the same wall. Hundreds of millions of products is more selection than a keyword search bar can reasonably help anyone navigate. His answer is conversational AI, shipping to customers as Alexa for Shopping.

What it actually does today. Type "where are my AA batteries?" into the Amazon search bar and you get an answer, not a page of battery listings: here are the ones you ordered, arriving today by 5 p.m.

Ask the difference between a drip machine and a cold brew maker and you get an explanation plus recommendations weighted by what it knows about you, like the fact that you live in a small apartment or have been browsing single-cup models. It reads images too.

Herrington photographed his fridge and got a weekly grocery list built around the inference that he cooks for two and prefers fresh food. Rajiv Mehta, VP of Conversational Shopping, said his son shot two bookshelves, about 200 books, and got reading recommendations that accounted for the NFL season starting.

Why sellers should care.

Mehta's framing is the tell: "Customers on Amazon know what to type and what not to type. We're about to open the aperture on what they can type."

Herrington's stated ambition is 10x the selection with no added friction, which means discovery gets decided by how well your product data answers a question rather than how well your title matches a phrase.

Herrington's five-year prediction: your house reordering its own supplies will feel completely normal. "Right now it seems like science fiction."

Productrise ran the same product-centric queries through standard Google search and Google's AI Mode on the same day, for 21 days in July 2026: more than 2 million listings across 100,000+ SERPs and AI Mode responses in the US and UK. The gap is enormous.

The numbers:

That last figure is the one sellers should sit with: ranking well in Shopping carousels does not carry over to AI Mode. The two surfaces are picking almost entirely different products. A search for "neck cream" returns a full shoppable carousel in standard search and a block of text linking to reviews in AI Mode, with zero buyable listings.

Why it matters now. AI Mode opens straight from the normal search box, so nobody has to change how they search for this to become the default. If it does, most of the organic Shopping visibility brands spent years earning evaporates overnight.

What to do about it. A standard SERP can hold up to 60 products, so mediocre listings still slip onto page one. AI Mode has no room for that. It's quality over quantity, which means optimized titles and descriptions across your entire catalog, not just hero SKUs, plus Google's new conversational attributes, which most brands are ignoring.

The counterargument. Huijer's own read is that fewer slots may mean each one is worth more: being one of four recommendations beats being one of twenty, and AI-sourced traffic has been converting above average.

Callum Lockwood of Re:Signal adds that AI shopping is conversational, so a brand gets multiple shots across a journey rather than one, and notes ChatGPT already pulls from Google's Shopping Graph, meaning Shopping data optimization now feeds the wider AI ecosystem.

Jason Berkowitz of Break The Web wonders whether the scarcity is permanent or just a stage before Google expands monetization and organic slots together.

The visibility pie is shrinking. Whether each remaining slice is worth more is the open question, but either way, your product data is what decides if you're in one.

No time to watch a webinar replay?

Download the summary:
"You're sitting on your most profitable channel — and starving it."

Kevin King and Norm Farrar ran a 90-minute working session on email for e-commerce sellers week before last, and this is the full write-up. The core argument: Amazon, Meta, Google and TikTok are rented land, ad costs up ~48% since 2024, AI overviews eating ~60% of clicks, only ~2% of your social followers ever seeing a post.

Email is the one channel no algorithm can evict you from, it drives 23% of all online orders, and one subscriber is worth about $45 over their lifetime if you actually treat the list like an asset.

Inside you'll find the five gates every email must pass (inbox, open, read, click, convert) and exactly which two most sellers are quietly bleeding out at, plus the send-frequency table by segment, the real numbers behind a $64,000 two-hour send, and a side-by-side of the same product written as AI prompt-slop versus a real angle — same AI, completely different result.

The back half gets practical: what the Dragonfish system actually does (24 client-specific brains, Stanford STORM research, separate warm and cold infrastructure, IP resolution that identifies 40–70% of your anonymous site visitors), what it costs and the margin math on whether it's worth it for your product, and Kevin's honest "not a fit if…" list.

Best of all, it includes all four copy-paste prompts he gave away free: Customer Language Miner, Subject Lines First, Objection Finder, and Winner Pattern Extractor that you can run tonight without buying anything, plus the weekend homework list.

Grab it, run the prompts, and check your SPF, DKIM and DMARC while you're at it.

Package design expert Kitty Lai says Shoppers notice your icons before they read a single word of your copy, and those tiny visual cues often decide whether they keep looking. Icons aren't decoration or a trend chase. They're a comprehension shortcut.

When a customer is juggling a dozen distractions and has about four seconds of patience left, every element on your packaging and listing images needs to help them recognize, understand and trust your product fast.

Three rules for getting more across in less time:

Keep it functional. Use icons to flag the things that actually drive the buy, like vegan, recycled, Made in UK, dishwasher safe. Customers shouldn't have to hunt for the important bits.

Match your brand personality. A playful brand can pull off a loose icon or even an emoji. A premium brand is better served by clean, understated marks that let the restraint do the talking.

Make it instantly recognizable. Run the same visual language across packaging, your website, your ads and your product pages. Familiarity is what makes customers spot you in a crowded search result and trust you when they land.

The goal is less clutter, more clarity. The right visual cues tell people what you're offering before they've read a sentence, which cuts hesitation, builds confidence and makes the decision feel easy.

Because the strongest packaging doesn't explain more. It communicates faster.

✌🏼 See you again Thursday …

The answer to today’s STUMP BEZOS is
The average half-life of a Youtube video is 3-4 weeks