Traffic But No Sales: How to Diagnose an Amazon Listing in the Right Order

A listing with traffic and no sales gets one of two treatments: the seller rewrites the bullets, or the seller redesigns the images. Both are guesses. There is an order to this, and images sit about two thirds of the way down it — not because images do not matter, but because five cheaper checks sit above them, and three of those can make a perfect image set irrelevant. Here is the order, what to look at in Seller Central at each step, and the specific signature that says the images really are the problem.

Step 0: is the sample big enough to mean anything?

Open Business Reports, "Detail Page Sales and Traffic by Child Item", and set the window to 30 days. You want two columns: Sessions and Unit Session Percentage. Sessions are visits, not page views; Unit Session Percentage is Amazon's own conversion figure for the listing.

Thirty sessions and no orders is not a diagnosis, it is noise. Three hundred sessions and no orders is a signal. Before that threshold, everything below this line is speculation you will pay for in redesign time.

What counts as a bad rate is harder to state honestly than most articles admit: Amazon does not publish conversion benchmarks by category. The widely quoted seller-side rule of thumb, published by My Amazon Guy, is that under 8% is poor, 8–12% is average and 13–15% is strong. Treat that as a practitioner benchmark rather than a platform figure, and compare mainly against your own listing's history and your own other SKUs.

Step 1: are you actually the Featured Offer?

This is the check that costs sixty seconds and cancels the rest of the exercise more often than anything else. If you do not hold the Featured Offer, the yellow Add to Cart button is not yours. Shoppers who land on your detail page buy from whoever holds it, and your session count looks healthy while your unit count does not.

The Buy Box percentage column sits in the same Business Report. If it is low, stop reading about images and go and look at your price, your fulfilment method and your competing offers on the same ASIN.

One 2026 wrinkle worth knowing: from July 2026 Amazon began removing the standalone seller-performance eligibility gate for the Featured Offer, folding those metrics into a single ranking formula alongside price, delivery speed and service quality. The rollout started in the US in early July and reached the EU and UK on 20 July 2026, with global completion expected by the end of the year. The practical effect is a wider pool of offers competing for the same button, so a Buy Box share that quietly slipped this summer may have nothing to do with anything you changed.

Step 2: find where in the funnel they leave

If you are brand registered, the Search Query Performance report in Brand Analytics is the single most useful screen on this problem, because it splits the funnel into impressions, clicks, cart adds and purchases, and shows your brand's share of each against the whole query.

Read it as a set of ratios, not as absolute numbers:

What the funnel shows	What it points at
Impression share healthy, click share far lower	The search grid: main image, title, price, star rating, delivery promise. Nothing inside the listing.
Click share healthy, cart-add share far lower	The detail page: gallery, price against what the shopper just saw, reviews, variation confusion.
Cart-add share healthy, purchase share lower	Checkout-stage friction: delivery date, shipping cost, an add-on or minimum-order flag.
Everything low, across every query	Relevance. You are being shown for searches your product does not answer.

Without Brand Registry you get a coarser version of the same read from advertising: the search term report shows clicks and orders per term, and a term with many clicks and no orders is either the wrong term or a page that loses people once they arrive.

Step 3: price, delivery and the offer itself

Open your own listing in an incognito window and look at it as a comparison, because that is what the shopper does. Your price against the three offers above you in search. Your delivery date against theirs. Whether they have a coupon badge and you do not. A listing 20% above the category and two days slower does not have an image problem; it has an offer problem wearing an image problem's clothes.

Step 4: the trust floor

Rating and review count act as a gate rather than a slider. Below roughly four stars, or in single-digit review counts against competitors with hundreds, shoppers filter you out before they read anything you wrote or look at anything you designed. No image set recovers a 3.4-star average, and money spent on design at that point is money spent decorating the wrong problem. Read the one and two-star reviews instead: they usually name the fix.

Step 5: now the images, in this order

By this point you have earned the right to blame the gallery. Work through it in the order the shopper meets it:

The main image, judged at thumbnail size. Its only job is the click, and it competes in a grid. Pure white background at RGB 255,255,255, product filling roughly 85% of the frame, no text, logos, watermarks, borders or props. Shrink it to 200 px on your screen: if you cannot tell what it is, neither can the search grid.
Slot two, the single strongest reason to buy. Not a feature list. One claim, legible on a phone.
The question your reviews keep asking. Pull the last fifty reviews and the customer questions. Whatever people ask twice belongs in an image, because a question that reaches the Q&A section is a purchase that did not happen for everyone too impatient to ask.
Dimensions and what is in the box. In most physical categories these two prevent more lost sales and more returns than any lifestyle shot.
Mobile legibility across all of them. Sources disagree substantially on Amazon's mobile share — figures from 60% to 75% of visits circulate, and at least one survey still puts the majority of completed purchases on desktop — so treat the exact number as unsettled and the design implication as not: a meaningful share of your buyers reads these images at a few centimetres wide.

The signature that images really are the bottleneck: click share is fine, cart-add share falls off, price and rating are competitive, and the top three listings in your category all carry dense, well-argued infographics while yours carries five product angles on white. That combination is worth redesigning for. Any single one of those on its own is not.

Step 6: change one thing, then measure it

If you are brand registered, Manage Your Experiments will A/B test main images, titles, bullets and A+ content against live traffic. Amazon does not publish a minimum session count, but low-traffic ASINs simply cannot be selected, and Amazon's guidance is to expect reliable results in the eight-to-ten week range. That is a long time, which is exactly why the order above matters: you do not want to spend ten weeks testing a gallery on a listing that was losing the Featured Offer the whole time.

If you cannot run an experiment, change one variable, note the date, and read Unit Session Percentage over the following four weeks against the four before. Changing the images, the title and the price on the same afternoon guarantees you will never know which one worked.

What this order deliberately puts last

A+ Content is worth having and worth doing well, and Amazon's own published figures put the lift at roughly 8% for basic A+ and up to about 20% for premium A+. But it sits below the fold, so it only ever converts shoppers who already got past the price, the rating and the gallery. If people are leaving above it, improving it changes nothing. Same reasoning for backend keywords, brand story modules and video: real, worth doing, and useless as a first move on a listing that is losing people in the search grid.