Why Your Amazon Listing Gets Traffic But Doesn't Convert

9 August 2026 · 9 min read · AMZNGO

Sessions look healthy. The keyword is ranking. Ad spend is going out and clicks are coming back. And the units still aren't moving. At that point the problem has stopped being a traffic problem and become a conversion problem, and the two are fixed in completely different places.

Below are the seven causes we run into most often on client accounts, ordered roughly by how much money they tend to cost. It is worth working through them in that order rather than starting with whichever one is easiest to fix.

1. The main image doesn't survive being a thumbnail

Almost every decision about your listing is made before anyone sees your listing. On a search results page your product is competing as a small square against a dozen others, and most of that traffic is on a phone. A main image that looks considered at full size can be completely unreadable at the size it is actually judged at.

The test is simple and slightly brutal: shrink your main image to about the size of a postage stamp, put it in a row next to your four closest competitors, and ask someone who has never seen the product which one they would tap. If yours needs explaining, it has already lost.

The usual failures are the product floating too small in the frame, a colour that disappears against white, and detail that only reads at 1000px. Filling the frame, choosing the angle that makes the silhouette recognisable, and checking contrast at thumbnail size fixes most of it.

2. The image stack repeats instead of building an argument

A listing gallery has a job: take someone from "this might be it" to "yes, this one" without them leaving to research. That is a sequence, not a collection.

What we see instead is six photographs of the same object from angles that add nothing — front, slightly-left front, slightly-right front. Each slot should be answering the next question the buyer has:

• What is it, exactly? The clean product shot.
• How big is it? Scale against something universally understood, or dimensions on the image.
• What makes it better than the cheaper one? The one feature worth paying for, shown rather than claimed.
• What's in the box? Everything laid out, so nobody has to guess whether the mount is included.
• What does it look like in my life? The lifestyle shot, in a setting your buyer recognises as theirs.
• Why should I trust this? Materials, certifications, warranty, guarantee.

Write the question above each slot before designing anything. If a slot has no question, it is a wasted slot.

3. The listing never handles the objection that stops the purchase

Every category has one question that quietly kills sales. For kitchen equipment it is whether it fits the space. For anything electronic it is compatibility. For apparel it is sizing, for supplements it is what is actually in it, for furniture it is how hard assembly really is.

You already know what yours is, because it is the question in your customer messages and the reason in your returns. Find the three most common complaints in your own reviews and the reviews of the two competitors above you, and check whether your listing answers all of them before the buyer has to scroll. If the answer is buried in bullet five, it is not answered.

4. Mobile is the real listing and it is being designed second

The majority of Amazon browsing happens in the mobile app, and the mobile layout is not a narrower version of the desktop one. Images are cropped differently, the bullet list is truncated behind a tap, A+ content modules stack and reflow, and text that was comfortable on a monitor becomes unreadable.

Design decisions that only work on desktop — small annotation text on infographics, wide side-by-side comparisons, anything relying on a caption under an image — quietly fail for most of your traffic. Check every asset on an actual phone, in the app, before it ships. Not in a browser window that has been made narrow.

5. Price is being read without context

Buyers do not evaluate price, they evaluate value, and they do it by comparing whatever is on screen. If you are more expensive than the listing next to you and nothing on your listing explains why, you are simply the expensive one.

The fix is usually not discounting. It is making the comparison legible: quantity or duration of use, material grade, what is included that the cheaper option leaves out. A two-pack at a higher number can convert better than a single at a lower one, purely because the unit maths is now visible.

6. The traffic is the wrong traffic

This is the one people miss, because the symptom looks identical: sessions up, conversion down. But if broad-match campaigns or a loosely relevant keyword are bringing in people looking for a different product, no amount of creative work will convert them, and the listing gets blamed for a targeting problem.

Pull your search term report and check conversion by term rather than in aggregate. If a handful of terms are eating spend at a conversion rate far below the rest, the listing is fine and the campaign needs work. It is worth ruling this out before commissioning any design.

7. Social proof gap against everything on the same screen

Reviews are the one element of your listing you cannot design your way out of. If the three listings around you have hundreds of reviews and you have a handful, a share of buyers will choose them regardless of how good your images are.

That does not make the creative work pointless — it makes it more important, because it has to carry more. But it does mean the honest plan includes actively building review volume through legitimate routes: Amazon's own Request a Review, the Vine programme where you are eligible, and inserts that ask for feedback within Amazon's rules. Anything that involves incentivising or filtering reviews risks the account, and is not worth it.

How to work through it

In order: rule out the traffic-quality problem first because it is free to check, then fix the main image, then rebuild the image stack around real buyer questions, then handle the objection, then re-check everything on a phone. Change one thing at a time and give each change enough time to produce a readable result — Amazon's data is noisy, and two changes at once tells you nothing about which one worked.