What changes when AI sits between your listing and the buyer | 80/20 of Ecom [5/22/26]

The sellers I talk to aren't short on data. They're short on signal.

This week we learned Amazon moved AI into the search bar. 

Tariffs are stacking up in ways most sellers aren't tracking. 

And the platform keeps adding layers faster than any team can watch by hand.

The operators pulling ahead aren't reading more dashboards. 

They're building workflows that watch the noise and hand them the one decision that matters.

Here's the 80/20 of Ecom for this week: 

In today's issue:

Let's get into it.

Amazon put its AI shopping assistant right where buyers start: the search bar.

Alexa for Shopping can now answer questions directly inside Amazon search. It pulls product comparisons from your results. It generates AI overviews on product detail pages. It shows up to a full year of price history. And it can run recurring shopping tasks through Scheduled Actions.

This isn't another standalone assistant feature you can ignore. Amazon is putting AI in the exact spot where customers search, compare, and decide.

The search bar is turning into a buyer's advisor.

Here's what that means for you. The question used to be "how do I rank?" Now there's a second question sitting on top of it: "does my listing actually answer what the AI is being asked?" Your title, bullets, A+ content, and backend data all feed what that assistant can pull and show.

Takeaway: Pull up your top 5 ASINs this week. Ask yourself what a buyer's AI assistant would surface from each one. If the answer is "not much," your listing work just got a new benchmark to hit.

Read it here: https://www.aboutamazon.com/news/retail/how-to-use-amazon-shopping-ai-assistant

Amazon pulled freight, distribution, fulfillment, and parcel services under one roof and called it Amazon Supply Chain Services. Then it opened the whole thing to businesses beyond marketplace sellers.

In plain terms: Amazon is now renting out the same logistics machine it uses internally. To anyone. Not just people selling on Amazon.

The signal is bigger than the branding. Amazon is turning its spare logistics capacity into a product. That gives it more data, more revenue, and more control over how goods move across the whole retail economy.

For you, it cuts both ways. More logistics options can mean more efficiency and lower cost. It can also mean deeper dependency on infrastructure Amazon controls. And that dependency gets harder to walk back the further in you go.

Takeaway: If you're eyeing Amazon Supply Chain Services, run the math on both sides. Not just the fulfillment cost today, but what flexibility you give up if Amazon changes terms or pricing down the road.

Read it here: https://www.freightwaves.com/news/amazon-rebrands-third-party-logistics-arms-as-unified-supply-chain-service

Here's the one that quietly eats margin.

FreightWaves reports that stacked duties on some product lines now land in the 20% to 80% range. That's from an Infios analysis of millions of U.S. customs entries.

The finding that matters: importers have stopped just accepting the duty bill. They're redesigning around it. Changing how they pick shipping modes, where they warehouse, how they structure customs entries, and how they plan landed cost. Tariffs are now a core planning input, sitting right next to freight cost, lead time, and service level.

This is the part most sellers miss. A tariff increase is easy to see. What's hard to see is how stacking across multiple categories compounds into a margin leak that doesn't show up until you're halfway through a bad quarter.

Takeaway: If you haven't rebuilt your landed-cost model in the last 90 days to account for stacked duties, do it this week. Not next sourcing cycle. Now. The sellers designing around tariffs are finding margin. The ones just tracking them are losing it.

Read it here: https://www.freightwaves.com/news/tariff-stacking-is-rewiring-supply-chain-execution

Quick gut check. No lecture.

Level 1 — Prompting. You download a report, paste it into Claude, ask a question, copy the answer. Useful. But it's manual, it's one-off, and it resets to zero every week.

Level 2 — Workflows. The workflow pulls the data, analyzes it, surfaces the exceptions, and hands you the short list. This is where real leverage starts.

Level 3 — Agent operating system. Multiple workflows running across PPC, inventory, Buy Box, and listings — with you approving the calls at each step. This is where the serious 7- and 8-figure operators are heading.

Most sellers are stuck at Level 1 and think the fix is a better prompt.

The fix is a workflow.

Here's what that looks like in real life. One student in our OpenClaw AI Ecom Accelerator built a workflow to monitor 450 SKUs across 5 marketplaces — Buy Box ownership, review counts, star ratings, and inventory risk. It delivers a structured exception report straight to Telegram. Nothing happens automatically. Her team reads the short list and decides what to act on.

Another student went deep on PPC. He built a workflow that pulled search analytics, search category, search performance, top search terms, and sales and traffic data — connecting signals that no single report shows on its own. The dataset was too big to review by hand. The workflow made it usable.

That's the jump from assistant to analyst. And it's exactly what I'm demoing live next week.

Quick question 

Which level are you on the AI curve? 

L1, L2, or L3? Hit reply and let me know.

I read every reply, and it shapes what I demo live on 5/27.

I'm running a live session with MerchantSpring to show what this actually looks like in practice.

This is not a "100 prompts for Amazon sellers" session. It's the shift from download report → paste into Claude → copy answer to data source → AI workflow → exception report → human decision.

What I'll cover:

Who it's for: Amazon brands, sellers, agencies, PPC managers, and ecommerce operators running a serious business.

When:

It's free.

👉 Save your spot: https://newsletter.7figuresellersummit.com/merchantspring-masterclass

And if you missed it: I posted a walkthrough showing how I used AI image generation, OpenClaw, Codex, and PickFu human validation to test an Amazon main image against a top competitor. The point is simple — AI shouldn't just generate ideas. It should run a feedback loop.

Watch it here: https://youtu.be/7eK3Hi_V9lk

Talk soon,

Gary

PS:  If your current “AI workflow” still means downloading reports, pasting them into Claude/ChatGPT and copying the answer back into a spreadsheet, this masterclass is for you. RSVP Here