People don't buy products on Amazon. They buy photos. That was Kevin's opener for a recent BDSC panel with Matt Kostan of ProductPinion, Kamal Singh of AMZ One Step, and John Aspinall. The three of them spent the next 90 minutes on the one image that decides whether anyone clicks at all.

Kamal's number is the reason to act this week. Across the tests his team runs, when a seller split-tests the main image they already have, the challenger beats it about 80% of the time.

The panel's deadline: retest at least every three months and look at your search results page every week, because a static main image dies quietly. Your CTR drifts down while competitors reshoot theirs, and nothing in Seller Central tells you.

Test angles before variations

This is where most sellers waste a month. They test five tiny tweaks against each other (badge on the left, badge on the right, slightly warmer white) and see nothing move.

Kamal splits it into two kinds of tests. Angles are fundamentally different concepts: packaging forward, hang tag, a person using it, a visual size reference, a dramatic demonstration. High risk, high reward, and the only place you find a 2 to 3x lift.

Variations are tweaks inside the winning angle: the color of the made-in-USA flag, different copy on the same box. Smaller lifts.

So run the angle test first. Five different concepts, one round. Then spend the following rounds refining whichever one won.

Kamal's sanity check before any of it goes live, the 5-2 rule: show the image to five strangers for two seconds. If they can't say what the product is, you decorated past the point of usefulness.

Guidelines are not terms of service

John was blunt. Nobody loses an account over a hero image. The worst outcome is suppression, and you swap the old one back and you're live in 15 minutes. The white-background rule is loosely enforced outside a few categories like shoes and jewelry. Bed sheets, supplements, desks are open territory.

The suppression dodge: upload the new image to a dummy listing first. If the bot accepts it there, push it live on the real one. Sometimes a few pixels of movement on a badge is what gets it through.

Test it the right way

Most sellers run A vs. B vs. C in isolation and pick the vote winner. That doesn't predict Amazon.

Matt's sequence: first, baseline. Drop your current image into a simulated search results page against your actual competitors and see where you rank today.

Second, test the new concept in that same environment, because the question is whether it wins on the shelf, never whether it wins alone.

Third, use 75% statistical significance as your green light. Ninety-nine percent is for science. Fourth, watch what people do over what they say. A 54/46 vote split is a coin flip, and when money is on the line the "loser" often wins.

Rule of thumb for sample size: 50 responses per option. ProductPinion runs the simulated-SERP tests and gives 20 free shoppers on any question to start. PickFu works too. Kamal's ListingOptimization.ai generates the variants and now pushes them into ProductPinion for one-click testing.

Kamal's cheapest conversion lift for the stack: one image with a concrete visual reference instead of a number. A pizza delivery bag listed at 24 inches tall means nothing. The bag shown holding six extra-large pizza boxes is understood instantly. Slot it around image four or five. He's seen 2 to 3x conversion lifts from that single image.

Pick your five angles this week. Run the test. Eight out of ten times you'll find something better than what's up there now.

The full panel, including Matt and John's Big Book of Main Image Ideas (52 tactics), the image-recognition tool Matt gives away free, and the AI side of images we left out here, is inside the Billion Dollar Sellers Club. Try it for $9 at billiondollarsellers.club.

4 things to fix before Q4:

Bonus: why packaging photos win the main-image slot

Not really about the on-box text. It's pixel utilization. A packaged product fills the entire image frame (no white space, no floating product on empty background), so it visually dominates more real estate in the search results grid and wins the scroll before anyone reads a word.

On-pack text still matters, but it closes the sale. The pixels get the click. Takeaway: treat the main image as territory to fully occupy, not just a photo to look nice.

Container rates keep climbing. Drewry's World Container Index shows Shanghai to New York spot rates jumped 7% last week to $10,394 for a 40-foot container, nearly triple what it cost a year ago.

The numbers:

Why it's happening: carriers are running blank sailings to manage capacity ahead of China's Golden Week, nine are scheduled for next week versus eight this week. Drewry expects Transpacific rates to keep climbing into the holiday, while Asia-Europe rates soften as Suez traffic gradually returns and demand stays weak.

Bottom line for sellers: if you're importing from China to the East Coast, budget for rates to stay elevated through Golden Week. West Coast and Europe lanes are more mixed, but capacity is tight across the board heading into peak season.

✌🏼 See you again Thursday …

The answer to today’s STUMP BEZOS is
Subscribe & Save drives a 1.8X sales conversion increase