Which listing changes actually move profit, and which just move traffic?

Most listing work increases impressions without increasing profit, because the changes target discovery rather than conversion, and conversion is what lowers your cost per order. Adding keywords, expanding backend terms and rewriting bullets for search all buy you more traffic at the same conversion rate, which means more ad spend for the same contribution. Changes that lift conversion rate do the opposite: they cut cost per order on every click you were already buying, and they compound with advertising rather than competing with it.

So the first question on any listing task is simple. Does this make more people see the page, or does it make more of the people already on the page buy?

What is the difference between ranking work and conversion work?

Ranking work changes what queries you appear for: title terms, backend search terms, structured attributes, category and browse node placement, variation structure. It sets the size of the pool.

Conversion work changes what happens once someone lands: main image, image sequence, the first two bullets, price and coupon presentation, review count and rating, A+ content, and the answers to the objections the reviews keep raising. It sets the rate at which the pool turns into orders.

Both matter, but they do not pay back the same way. A 10% lift in impressions at flat conversion gives you 10% more sales at the same cost per order. A 10% lift in conversion rate gives you 10% more sales and a 9% lower cost per order, which lands directly in Keep Rate. If you only have capacity for one, take conversion.

Ranking work buys you more clicks. Conversion work makes the clicks you already bought cheaper.

Why is the main image the highest leverage asset on the page?

The main image is the only element that works in three places at once: search results, sponsored placements, and the detail page. It decides your click through rate in search, which decides your cost per click in advertising, which decides your cost per order. A better main image lowers what you pay for every click before anyone reads a bullet.

The rules are unglamorous. Fill the frame, so the product is not floating in white space at thumbnail size. Make the format legible at 200 pixels wide, because that is the real viewing size. Show quantity and size when the category shops on it, within Amazon's image policy. Test one variable at a time and give each test enough sessions to mean something.

We have seen a main image swap move conversion from 11.4% to 14.9% on a mature product with no other change. That is a full point and a half of Keep Rate on a product line, from one asset.

What is A+ content actually worth?

Less than agencies imply, and more than nothing. On products where the objection is technical, sizing based, or comparison driven, good A+ content reliably adds one to three points of conversion, because it answers the question that was stopping the purchase. On simple, low consideration products it adds close to nothing, because nobody scrolls that far.

The comparison module is the exception worth building everywhere. It keeps a shopper inside your catalogue instead of sending them back to search, and on brands with three or more related products it is usually the single best performing module on the page.

How do variations quietly cannibalise each other?

Variations are sold as free review consolidation. What they also do is let shoppers pick the cheapest child, and put your worst margin child in front of demand that would have bought the better one.

Check three things on every parent: which child holds the buy box on the parent page, whether the price ladder rewards trading up, and whether a low margin child is absorbing advertising meant for the range. This is the same per SKU discipline described in the Keep Rate breakdown, applied one level down.

What is the honest role of keywords in 2026?

Keywords still decide eligibility. If a term is nowhere in your listing or structured data, you will struggle to rank organically for it. That is the whole of their power now.

What they no longer do is rank you by repetition. Stuffing terms into bullets costs you readability, which costs conversion, which raises cost per order. Cover each relevant term once in the most natural place, fill the structured attributes properly because Amazon's own retrieval leans on them heavily, and spend the remaining effort on copy that answers objections. Indexing is a checkbox, not a strategy.

What is the prioritised checklist?

1. 01Fix anything suppressed, out of stock or missing the buy box. No optimisation matters if the page cannot sell.
2. 02Main image. Test it, measure click through rate in search, keep the winner.
3. 03First two bullets rewritten against the top three objections in recent reviews.
4. 04Image sequence: use, scale, ingredients or specification, comparison, social proof.
5. 05Price and coupon presentation, checked against the variation ladder.
6. 06A+ content, comparison module first.
7. 07Structured attributes filled completely, including the ones buyers filter on.
8. 08Keyword coverage, once each, in natural language.

Work top down and stop when the return goes flat. Then measure it properly, because a conversion lift shows up in advertising cost before it shows up anywhere else, which is exactly the failure mode ACoS hides.