Listing & Conversion

Why the Main Image Decides Your CTR

The main image is the single most-seen element of your listing. It appears in every search result, every sponsored placement, every browse page — and it is what a shopper judges before reading a single word. That makes it the highest-leverage thing on the listing. Improve it and everything downstream improves: more clicks for the same impressions, better conversion, better organic rank. Watch the walkthrough Prefer to watch? One hundred and seventeen seconds on why the top of the funnel multiplies through everything after it, how to tell whether you have a problem, and how to size the prize. Why it moves the needle more than anything else Amazon is a scanned interface. A shopper works down a grid of near-identical thumbnails, and most of them get a fraction of a second. Everything else in your listing only matters after the click. Bullets, A+ content, reviews, price presentation — none of it is seen by someone who scrolled past your image. So the main image sits at the top of the funnel, and a percentage gained there multiplies through every stage after it. A 15% CTR improvement is 15% more traffic at no additional advertising cost, on every keyword you appear for, permanently. First: work out whether you have a problem Do not start a design project on instinct. Establish where your click-through rate actually sits relative to the market. Search Query Performance gives you both halves of the comparison: 1. Your ASIN's CTR on the top five keywords in your niche. 2. The average CTR of those same five keywords across all sellers. 3. Compare them. Your CTR against the keyword average If your CTR is below the average, this is urgent. Shoppers are seeing you and choosing someone else, and every advertising pound you spend is buying impressions that convert to clicks at a below-market rate. If your CTR is above average, it is still worth testing. Being ahead of the average is not the same as being at your ceiling, and this is the cheapest improvement available to a listing that is otherwise working. For reading the report properly, see Understanding Search Query Performance and Keyword Research from Search Query Performance. Size the prize before you commit The arithmetic is simple and worth doing, because it tells you how much effort is justified. Take your current impressions and CTR. Model the same impressions at a CTR one or two points higher. Convert the additional clicks at your current conversion rate, and multiply by margin. On a product with meaningful impression volume, a two-point CTR improvement is frequently worth more per year than any advertising optimisation you could run — and unlike advertising, it costs nothing ongoing once the image is live. That number is also what tells you whether to hire a good designer or a cheap one. The three phases | Phase | What happens | |---|---| | 1 | Gather 3–5 reference images, then write a brief for each | | 2 | Have a designer produce 2–3 new candidates | | 3 | Test them against each other, then live on Amazon | The structure exists to stop the two common failure modes: designing from your own taste rather than from what already works in the category, and swapping images without measuring whether the new one is better. What to expect Not every new image wins. That is why you produce two or three and test rather than commissioning one and hoping. A losing test is still information. It tells you the current image is stronger than you assumed, which is worth knowing before you spend further. Start with one or two products. Your highest-traffic listings, where a CTR gain is worth the most. Once the process works, run it monthly across more of the catalogue. Related - Finding Reference Images — phase 1, where the ideas come from. - Writing an Image Brief — turning an idea into something a designer can build. - Testing a New Main Image — proving it is actually better. - Preparing a Launch: CTR and CVR Tactics — where the main image fits in a relaunch. - Packaging Design for Amazon — the physical product behind the image.

Finding Reference Images

The worst way to design a main image is to imagine one. The best way is to find images that already work, understand why, and adapt them. This phase produces 3–5 reference images — real listings that look close to what you want — each with a written note on exactly what you are taking from it. Watch the walkthrough Prefer to watch? One hundred and twenty-five seconds on the five sources ranked by usefulness, what an image has to do in a fraction of a second, and why three directions beat three variations. Copy, adjust, paste The principle, stated plainly: you are not inventing a composition, you are adapting a proven one. A reference image is a photograph of an existing Amazon product that resembles the result you want. Alongside it you write what specifically you are borrowing — the angle, the background, the way a benefit is called out, the lifestyle context, the way scale is communicated. Without that note the reference is useless, because a designer will copy the wrong thing. "Like this one" is not a brief. Where your clicks come from, keyword by keyword Where to look Five sources, in roughly descending order of usefulness. 1. Direct competitors on your main keyword Search the term you most want to rank for and study the images taking the clicks. These are directly relevant, because they are competing for the same shopper in the same context. Look at the ones performing well, not just the ones ranking well. Brand Analytics shows which listings receive the most clicks for a keyword, which is a better signal than position alone. 2. Direct competitors in your subcategory Broader than one keyword, and it surfaces conventions you may be breaking without realising — if every listing in the category shows the product in use and yours is on white, that is worth knowing. 3. Adjacent categories The most productive source, and the one people skip. Look at products that are similar in kind but not competitors. Selling a toothbrush? Study electric toothbrushes. A reading pillow? Study pregnancy pillows. These categories have solved the same visual problems — showing scale, showing texture, showing use — without being your direct competition, so borrowing from them makes you distinctive rather than derivative. 4. Design and craft platforms Search your main keyword on visual and handmade marketplaces. The photography conventions there are different from Amazon's, and importing one is an easy way to stand out in a grid of identical white backgrounds. 5. Your own reviews Not a source of compositions, but of subjects. What customers repeatedly praise or complain about tells you what the image needs to communicate. What you are looking for An image has to do work in a fraction of a second. Aim for reference images that show one of: The main benefit, visible without reading. How the product makes someone feel — freshness, warmth, calm, cleanliness. Emotional register is doing more work than most sellers credit. The key feature, especially anything that distinguishes you from the listings either side of you. Scale, if size is a common question or complaint. And the single most valuable subject: whatever your competitors' 1–3 star reviews complain about. If the category's recurring complaint is that the product is smaller than expected, an image that settles size immediately is worth more than any aesthetic improvement. Building the collection Collect every candidate, without filtering as you go. Judgement comes later; the goal now is volume and range. For each one, write: what you are borrowing, and why you think it would work for your product. Then narrow to three. You want three distinct directions rather than three variations of one idea — if all three are the same composition with different backgrounds, your test will only tell you about backgrounds. Aim for genuinely different approaches: one that leads on the benefit, one that leads on use or lifestyle, one that leads on a feature or on scale. Then you learn something regardless of which wins. Check where your traffic actually comes from One practical decision belongs at the end of this phase, because it changes what the designer produces. Look at whether your sessions are mostly mobile or desktop, and design for the majority. The two crop differently and are read at very different sizes. An image with fine detail or small text may work on desktop and be illegible on a phone. See Understanding Business Reports. Common mistakes Designing from imagination rather than from what already works. Collecting references without notes, so nobody can tell what you meant. Three variations of one idea, which wastes the test. Only looking at direct competitors, and producing something that blends into the category. Ignoring the review complaints, which are the clearest brief you will ever get. Related - Why the Main Image Decides Your CTR — whether this project is worth running. - Writing an Image Brief — turning these references into instructions. - Testing a New Main Image — proving which one wins.

Writing an Image Brief

A designer cannot read your mind, and the cost of an unclear brief is a revision cycle — days lost and a worse final image, because each round drifts a little further from what you meant. Write one brief per image. Three images, three briefs. Watch the walkthrough Prefer to watch? One hundred and twenty-three seconds on what belongs in a brief, why the aspect ratio follows your traffic split, and the feedback discipline that avoids extra rounds. What a brief contains Your current main image. The starting point, so the designer can see what they are replacing and what changed. The reference images, with your notes on what specifically you are borrowing from each. This is the core of the document. Links to download the source files. Every product photograph, logo and asset the designer will need, in full resolution. A brief that makes someone ask for files has already cost you a day. A written description of the image you want, in as much detail as you can manage. The dimensions. Writing the description Be specific to the point of feeling excessive. Every ambiguity becomes a decision the designer makes for you. Describe what is in the frame and where. Product position, angle, what surrounds it, what the background is. Say what the focus is. What should a shopper notice first, in that fraction of a second? If everything is emphasised, nothing is. Describe the feeling. Clean and clinical, warm and domestic, energetic. Designers work well from this and badly from adjectives like "premium", which mean nothing concrete. Call out anything non-negotiable — the product must be unobstructed, a particular feature must be visible, certain text must appear. Paste the reference images inline, next to the paragraphs they relate to, rather than collecting them at the end. Proximity removes interpretation. Use annotated screenshots. Circle and label the exact element you mean on the reference. A marked-up image removes more ambiguity than a paragraph. Sessions and conversion, per product Choosing dimensions Determined by where your traffic comes from, which you established in the previous phase. | Traffic majority | Aspect | Pixels | |---|---|---| | Desktop | square | 3000 × 3000 | | Mobile | portrait | 2500 × 3000 or 2000 × 3000 | This is not a small detail. A taller image occupies more vertical space in a mobile results feed, which means more attention before the shopper scrolls. On a mobile-dominant listing, the aspect ratio alone can move click-through rate. Check the split before deciding rather than assuming — see Understanding Business Reports. Working with the designer Match the professional to the job. Standard graphic design, a 3D render and a photoshoot are three different skills, and a designer strong at one is often weak at another. Decide which you need before commissioning. Expect two or three rounds. The first batch is rarely right, and that is normal rather than a sign of a bad designer. Give feedback that is specific and comparative. Put the delivered image next to the reference and describe the difference concretely: "the product is too small relative to the frame — match the proportion in reference 2", not "it doesn't pop". Batch your feedback. Collect every note into one message per round rather than sending observations as they occur. Drip-fed feedback produces drift and irritation, and multiplies the number of rounds. Keep going until it is right. You are going to run this image on every impression the product receives for months. A further revision round is cheap by comparison. When you are done You should have two or three finished images, materially different from each other and from your current one, all built to the correct dimensions. That is the input for testing — see Testing a New Main Image. Common mistakes A one-paragraph brief, which guarantees revision rounds. References without annotation. Missing source files, stalling the job before it starts. Designing square by default when most of your traffic is mobile. Vague feedback that the designer cannot act on. Accepting something close enough because you are tired of the process. This image will be seen more than anything else you own. Related - Finding Reference Images — the inputs to this brief. - Testing a New Main Image — what happens to the finished images. - Why the Main Image Decides Your CTR — why this is worth the effort.

Testing a New Main Image

You have two or three finished images. The question now is which is best — and whether any of them beats what you already have. Two tests, in order. A cheap one that narrows the field, then a real one on Amazon that decides it. Watch the walkthrough Prefer to watch? One hundred and twenty seconds on the two tests and their order, how to read the result against a range, and the trap of judging on click-through alone. Test 1: pick the strongest candidate Before touching your live listing, run your new images against each other with a survey panel. Compare the new images to each other first, not against your current one. You are ranking them — best, second, worst — so that the image you put live is your strongest candidate rather than an arbitrary one. That single step meaningfully raises the odds the live test succeeds. Optionally, run your best new image against your top three competitors. Use Brand Analytics to find who actually receives the most clicks on your main keyword, and put your image alongside theirs. Reading that comparison: Placing first or second — proceed to the live test. Placing third or fourth — go back and improve before spending live traffic on it. This whole stage is optional and worth doing anyway. Panel results are not shopper behaviour — people asked to compare images behave differently from people shopping. But it is cheap and fast, and it stops you burning a live test on your weakest candidate. Test 2: live on Amazon The only test that counts. Upload the new image, then measure against a stable baseline. Watch Top of Search placements specifically. They are the most predictable and least noisy part of your traffic, which makes them the cleanest read on a CTR change. Pick your top three Sponsored Products campaigns by spend and track those. Placements, where the cleanest signal is Reading it, day by day then week by week Days 1–4 — check daily, but only for disasters. Sales falling clearly — the image is worse. Stop and switch to your second candidate. Do not wait for statistical certainty; you are losing money now. Sales stable or rising — leave it alone and let data accumulate. Four days is not enough to confirm a win, only to catch a failure. After one week — look properly. CTR, week by week over the last six weeks, from the Search Catalog Performance report. You are comparing the current week against the recent range rather than against a single previous week, because weekly CTR is noisy. CVR, from your business reports — unit session percentage. Advertising CTR and CVR on those three campaigns, Top of Search only. The verdict Compare this week's CTR against the range of the previous weeks: | Result | Reading | Action | |---|---|---| | Above the range | very good | keep it | | Top of the range | good | keep it, watch another week | | Bottom of the range | bad | consider the next image | | Below the range | very bad | switch to the next image | Then decide: Bad after one week — start the test with your second image. Middling after one week — give it one more week. Some of that is noise, and a second week usually resolves it. Good or excellent — keep it, and decide whether to stop or chase further gains with the remaining candidates. And check CVR alongside CTR. An image can raise clicks while lowering conversion — usually because it over-promises, and the shopper arrives to something different from what they expected. More traffic that converts worse is not a win, and CTR alone will not tell you. A note on risky images If your new image pushes against Amazon's image rules — heavy text, added graphics, unconventional backgrounds — there is a real risk of suppression, and getting your original back live can take time. For those, use Amazon's own experiment tooling rather than swapping the live image. It is slower, but a suppressed main image on a high-traffic listing is a serious outage. After a win Do it again. The first improvement is rarely the ceiling, and you now have a process and a designer who understands your product. Then move to your next product. Run this monthly across your catalogue, starting with the listings that get the most impressions. And re-check your CTR against the keyword average — that comparison is what started this project, and it is how you know whether you are now ahead of your category or merely ahead of where you were. See Understanding Search Query Performance. Common mistakes Testing the new image against the old one on a panel rather than ranking the new ones first. Switching after two days on normal fluctuation. Comparing to a single previous week instead of a range. Watching CTR and ignoring CVR. Reading all placements together, which buries the signal in noise. Stopping after one win. Related - Writing an Image Brief — producing the candidates. - Why the Main Image Decides Your CTR — the benchmark this is measured against. - Placements: Top of Search — why Top of Search is the cleanest read.

Why a Rating Drop Costs More Than It Looks

A rating falling from 4.3 to 4.2 sounds trivial. It is not, because of where the star display rounds. At 4.3 you show four and a half stars. At 4.2 you show four. Half a star disappears from every search result you appear in, and the shopper scanning that grid never sees the decimal. Watch the walkthrough Prefer to watch? One hundred and sixteen seconds on where the star display rounds, the four-stage cascade it triggers, and the order to work in. The cascade The half-star loss is not one problem, it is four, in sequence: Click-through rate falls. Your thumbnail now looks worse than the listings either side of it, on the one signal shoppers use most to compare. Conversion rate falls. Shoppers who do click see a lower rating on the detail page and hesitate. Organic rank falls, because CTR and CVR are what Amazon ranks on. Sales fall and TACoS rises, because you are now buying with advertising the traffic you previously got free. Each stage feeds the next, and the whole thing compounds quietly. A product can lose a third of its organic sales over a couple of months without any single alarming day. Conversion and sessions, per product This outranks whatever else you were doing While a leading product has a rating problem, other work on that product is largely wasted. Better advertising buys traffic that converts worse than it should. A relaunch pushes velocity into a listing shoppers are rejecting. A new main image lifts CTR into a lower conversion rate. Commit 3–5 hours a week to it until it is fixed. Not more — several of the fixes are slow by nature — but not less, and not intermittently. Short-term and long-term solutions do different jobs This is the distinction that makes the plan work. Short-term solutions buy time. They can move the displayed rating within days or weeks, but they do not change why customers are unhappy. Used alone, the rating drifts back down. Long-term solutions fix the cause. They take months, because they involve the product itself, your content, or both — but they are what makes the improvement permanent. Run both at once. The short-term work holds the line while the long-term work takes effect. Doing only the short-term is treating a symptom; doing only the long-term means bleeding sales for a quarter while you wait. The order to work in Day one: find out whether a variation is doing it. Removing or isolating a badly-rated child ASIN is the only fix that can work the same day, because parent ratings are blended. Check this first, in every marketplace. See Find the Variation Dragging Your Rating Down. Day one, in parallel: read the negative reviews properly and identify the top two or three recurring complaints. Everything else depends on knowing what customers are actually unhappy about. Then, immediately: order your competitors' products and get them to your supplier. This has the longest lead time of anything on the list, so it starts first even though it finishes last. See Fix the Product by Studying Competitors. While those are in transit: rewrite your content to answer the complaints. See Answer the Complaints in Your Content. Alongside all of it: run the short-term tactics in the order set out in Short-Term Rating Recovery. How hard to push On a leading product, use every tactic available. The cost of the rating staying down is larger than the cost of all of them combined. On a secondary product, or where sales have not moved much, be selective: fix the variations, run the automated recovery, and do the long-term work. Skip the expensive interventions. Watch the right numbers A rating problem shows up as conversion rate falling while sessions hold, then sessions falling as rank follows. Session and conversion history per product is where you see it — see Understanding Business Reports. Click-through rate against the keyword average confirms whether the star display is costing you clicks — see Understanding Search Query Performance. And set an alert on rating, because the drop happens gradually and nobody notices the day it crosses a rounding boundary. See Setting Up Smart Alerts. Related - Find the Variation Dragging Your Rating Down — the only same-day fix. - Short-Term Rating Recovery — buying time while the real fix lands. - Fix the Product by Studying Competitors — the highest-success long-term solution. - Dealing With Listing Hijackers — when the bad reviews are not about your product.

Find the Variation Dragging Your Rating Down

A parent listing shows one rating, but that number is a blend of every child ASIN underneath it. Which means a single bad variation can drag down a product family that is otherwise fine. This is the only rating fix that can work the same day, so it is always the first thing to check. Watch the walkthrough Prefer to watch? One hundred and twenty-four seconds on why blended averages hide the problem, the four decisions per variation, and the finding that matters as much as the fix. Why the blend hides the problem You see 4.2 on the parent. That average might be four strong variations at 4.6 and one at 2.9 — and if that one variation is a small share of sales, you may have no idea it exists. Common causes of a single bad variation: One size or colour with a genuine quality problem — a material that behaves differently at a larger size, a colour that fades. A variation with almost no reviews, where two or three bad ones set the average. A bundle or multipack where the value proposition does not land the way the single unit does. An old variation you stopped promoting but never removed, still accumulating occasional disappointed buyers. Product-level ratings and performance See the real numbers per child The problem is that the blended parent rating is what Amazon shows you most readily. To act, you need each child's own rating and review count. Temporarily un-merging the parent exposes the true rating and review count of every child ASIN individually. That is a catalogue operation and worth doing carefully — but it converts a blended average into real data. Once un-merged you can see: Each child's actual rating, not the family average. Each child's review count, which tells you how much weight it carries and how hard it would be to move. Which variations are pulling the parent down, unambiguously. Then decide, per variation Remove it — if a variation is badly rated, low volume and not strategically important, taking it out of the family lifts the parent immediately. This is the same-day fix. Isolate it — if it sells meaningfully but rates badly, separate it from the family so it stops contaminating the parent while you work on it. Fix it — if the problem is specific and solvable, and the variation matters, address the underlying cause and let reviews recover. Restructure — sometimes the honest answer is that the family was assembled badly, mixing products that should never have shared a listing. Then re-merge only the variations that support a strong parent rating. Check every marketplace Do this in each marketplace you sell in, not just your largest. Variation structures and their ratings differ between marketplaces, and a variation that is fine in one may be the problem in another. It is a quick check per marketplace and it is routinely skipped. See Marketplaces and Currency. A caution before acting Do not remove variations aggressively without seeing the un-merged data first. Removing a child ASIN loses its reviews from the family, and if you have misjudged which variation is the problem you will have thrown away reviews that were helping you. Blended averages are genuinely misleading in both directions. Look first. Then act. What you should end up with A clear list of which children rate well and which do not, with review counts attached. A decision recorded per variation — keep, remove, isolate or fix. A parent rating that reflects your good products, not your worst one. And critically: you now know whether you have a variation problem or a product problem. If every child rates badly, no amount of catalogue restructuring will help and the answer is in the product itself — see Fix the Product by Studying Competitors. That distinction is worth as much as the fix. Tracking the result Watch rating and conversion per child ASIN afterwards. If the parent rating recovers and conversion follows, the diagnosis was right. If the rating recovers and conversion does not, the complaint was about something the rating was only partly reflecting. See Sales Performance: Which Products Carry the Business and Understanding Business Reports. Related - Why a Rating Drop Costs More Than It Looks — what the half-star is worth. - Short-Term Rating Recovery — what to run once variations are sorted. - Answer the Complaints in Your Content — when the problem is the product family as a whole.

Short-Term Rating Recovery

Short-term rating tactics do not fix why customers are unhappy. They move the displayed number while the real fix is still months away — which matters, because every week at a lower star display costs sales, rank and advertising efficiency. Four tactics, in the order worth trying them. Watch the walkthrough Prefer to watch? One hundred and twenty-eight seconds on the four tactics in order of value, when Vine helps and when it backfires, and the honest limit of all of them. 1. Intercept one- and two-star reviews automatically The highest-value short-term tactic, because it works continuously without attention. Set up automation that, when a one- or two-star review appears: Issues a refund automatically. The customer already feels badly treated; a refund arriving unprompted changes that. Sends an apology and an offer to help. Not a request to change the review — that violates Amazon's policies — but a genuine offer to resolve the problem. A meaningful proportion of those customers update their review, because most negative reviews come from people who felt ignored rather than people who hate your product. Why this is first: it is cheap, it runs continuously, and it treats the cause of some negative reviews rather than only the symptom. It also generates useful information — the replies tell you what is actually going wrong. Conversion and sessions, watched through a recovery 2. Remove reviews that violate Amazon's policies A meaningful share of negative reviews are not about the product at all. They are about delivery time, packaging damage, courier behaviour, or the seller — and Amazon's own terms say product reviews should be about the product. Those reviews can be reported and removed, in the same way you would contest an unfair seller-account review. Look for reviews complaining about: Delivery speed or a late arrival. Damaged packaging where the product itself was fine. Amazon's service, or a courier. Something a different seller did, if the listing has other offers. A different product entirely, which happens more often than you would expect. Doing this at scale is a specialist job, and services exist that use automation to find qualifying reviews and process the removals. They are expensive — in the region of a couple of hundred dollars per review removed, and negotiable — so this makes sense on a leading product where the rating is costing you real money, and rarely on a secondary one. You can also do it yourself for the most damaging reviews. It is slow, but the cost is your time. 3. Use Vine on new ASINs Useful in one specific situation: your product normally rates well, but a batch of stock was worse and the reviews from that batch are dragging the average down. You cannot dilute an existing review pool quickly. But you can create a new one. The process: 1. Launch the product under new ASINs, or as new variations. 2. Send 35–50 units to cover the programme. 3. Enrol in Vine. 4. Wait until the units are claimed and you have 20+ reviews. 5. Compare the new rating to the parent's. If it is higher, add the new variation to the family or merge the ASINs. When this works: the underlying product is genuinely good and the bad reviews came from a specific, resolved cause. When it does not: the product itself has a problem. Vine reviewers will find it too, and you will have paid for a second set of bad reviews. Only worth doing if you have not already used Vine on the product, or if you have real reason to believe the new units will rate better. 4. Provide direct customer support Give customers a way to reach you before they reach the review box. Amazon's onsite support tooling lets you communicate directly with buyers, which matters most for products where the complaint is often misunderstanding rather than defect — assembly, setup, sizing, compatibility. A frustrated customer who gets a helpful answer within a day usually does not leave a one-star review. One who waits three days does. This is the slowest-acting of the four, and the one that most changes the long-run trend. Choosing which to run On a leading product where sales have visibly dropped: run all four. On a secondary product, or where the drop is modest: run the automation and the policy-violation removals, and put your effort into the long-term work instead. The expensive tactics are hard to justify. What these will not do None of these fixes the product. If customers are unhappy because the product genuinely disappoints, every one of these tactics is holding back water. The rating will drift down again as new reviews arrive. Run these to buy time. Run the long-term work to make it stick — see Fix the Product by Studying Competitors and Answer the Complaints in Your Content. Related - Why a Rating Drop Costs More Than It Looks — why this is urgent. - Find the Variation Dragging Your Rating Down — do this before any of the above. - Sales Performance: Which Products Carry the Business — watching rating and conversion recover.

Fix the Product by Studying Competitors

Of everything you can do about a rating problem, this has the highest chance of actually working — and it starts with buying your competitors' products. The logic is simple. Somebody in your category is selling a similar product with thousands of reviews at 4.6. Whatever they are doing physically differently is the answer, and it is knowable. Watch the walkthrough Prefer to watch? One hundred and thirty-five seconds on why this has the highest odds of working, what to ask your supplier to compare, and why you watch new reviews rather than the average. Start this first, finish it last It has the longest lead time of anything on the list. Buying competitor products, shipping them to your supplier, having them assessed, changing production, and manufacturing a new batch takes months. So it begins on day one, in parallel with the same-day variation fixes and the short-term tactics — even though its results arrive long after theirs. Pick the right competitors Find the products with the most reviews at the highest ratings. Not the best sellers necessarily — the best rated, with enough reviews that the rating is trustworthy. A product with 4,000 reviews at 4.6 has demonstrably solved the quality problem you have. A product with 40 reviews at 4.8 has not demonstrated anything yet. Buy at least three. One competitor tells you about one competitor. Three tell you what the category standard is, and where the consensus differences lie. Send them to your supplier Ship the competitor products directly to your manufacturer, and ask a specific question: what is different between these and what you make for us? This is the part that works, and it works because your supplier knows things you cannot see. They understand the material grades, the moulding, the stitching, the coating, the ingredient sourcing. Handed a physical competitor product, they will identify differences you would never find from photographs and reviews. Ask them to look at, specifically: Materials — grade, weight, composition. The single most common source of quality difference. Construction — how it is assembled, joined, finished. Packaging — which affects both damage in transit and the unboxing impression that colours a review. Size and weight — whether the competitor is subtly more substantial in a way customers feel. Taste, texture or smell, for consumables and anything applied to the body. Then change how yours is made Instruct the supplier to match the competitor. That instruction sounds blunt and it is deliberate. You are not asking for incremental improvement; you are asking them to reproduce the specification of a product that thousands of customers have rated well. This is the highest-probability fix available, because you are copying a proven outcome rather than guessing at one. The competitor has already run the experiment with real customers. Expect it to cost more per unit. Better materials usually do. Weigh that against what the rating drop is costing — and note that a higher unit cost you can price for is a much better problem than a rating that suppresses conversion, rank and advertising efficiency at once. Product performance, before and after a change Check the economics before committing A specification change alters your landed cost, which alters your margin and possibly your price. Work out the new cost per unit, and what it does to your margin at your current price. If the improved product needs a higher price, that is a decision to take deliberately rather than discover later. And update your sourcing cost with Amazon afterwards, because reimbursements are calculated from it — see Reimbursements, COGS and FBA Capacity. While you wait The competitor products take one to two weeks to reach an overseas supplier, and the new production run takes longer. Use that window for the content work — see Answer the Complaints in Your Content. It costs nothing, it takes about that long, and it starts working immediately. When the new stock lands Watch the rating on new reviews specifically, not the overall average. The overall figure moves slowly because it carries all the historical reviews; what tells you whether the fix worked is the rating of reviews arriving after the new batch reached customers. If new reviews are clearly better, the fix worked and the average will recover over time — and this is exactly the situation where Vine on a new ASIN becomes worthwhile, because you now genuinely have a better product. See Short-Term Rating Recovery. If new reviews are no better, the diagnosis was wrong. Go back to the complaints and check whether the issue was ever about the physical product — expectation-setting problems look identical in the rating and are solved entirely differently. Common mistakes Buying one competitor, and copying an idiosyncrasy rather than a standard. Choosing best-selling rather than best-rated competitors. Sending photographs instead of products. Your supplier needs to hold it. Asking for "better quality" rather than for a specific comparison against a specific product. Refusing the extra unit cost without comparing it to what the rating is costing. Waiting for this to finish before doing anything else. It runs alongside everything. Related - Answer the Complaints in Your Content — the work to do while this is in transit. - Why a Rating Drop Costs More Than It Looks — what justifies the cost. - Designing the Offer — the same supplier conversation, run before launch.

Answer the Complaints in Your Content

A significant share of negative reviews are not about a faulty product. They are about a product that was not what the customer expected. Those reviews are fixable without changing anything you manufacture — by making the listing set the right expectation before the purchase rather than after it. This is the cheapest long-term fix, it starts working immediately, and it fits neatly into the weeks you spend waiting for a production change. Watch the walkthrough Prefer to watch? One hundred and twenty-seven seconds on separating expectation complaints from quality complaints, why the answer goes in four places, and why specific beats reassuring. Start by reading the complaints properly Gather every negative review — one, two and three stars — across all your variations and marketplaces. Then group them by theme, and count. You are not looking for the worst review, you are looking for the most frequent complaint. A single furious review about something nobody else mentions is noise. Forty people saying the same thing is a brief. Rank the themes by frequency and take the top two or three. Those are what you address. An AI model handles this well. Paste in the reviews and ask for the recurring themes with counts. It is faster than reading several hundred by hand and less prone to your own bias about what you expect to find. Separate the two kinds of complaint Expectation complaints — "smaller than I thought", "the colour is different", "I did not realise it needed batteries", "it does not fit my model". These are content problems. The product performed as designed; the listing sold something slightly different. Quality complaints — "broke in two weeks", "arrived faulty", "material feels cheap". These are product problems, and content cannot fix them — see Fix the Product by Studying Competitors. Most listings have both, and the split matters because it decides where your effort goes. Sorting the themes into these two buckets is the single most useful thing you do in this exercise. Sessions and conversion, per product Answer each theme in several places For each of your top complaints, address it in every element a shopper actually looks at — because they do not read the whole listing, they sample it. One image dedicated to the main problem. If size is the complaint, an image that establishes scale unambiguously — the product next to a familiar object, or in use. Images do more work than text. One full A+ section. Enough room to explain properly, with pictures. One bullet point. Written to answer the objection directly rather than to praise the product. In the video, if you have one. Say it aloud. Video is where a shopper resolves final doubt. The repetition is deliberate. Different shoppers read different parts, and a complaint that appears in one place will be missed by most. Write to the objection, not around it The instinct is to be reassuring and vague. Be specific instead. Weak: "generously sized". Strong: "14cm tall — about the height of a standard mug", with an image showing exactly that. Weak: "easy assembly". Strong: "assembles in about four minutes with the included tool; no drilling". Naming a limitation directly costs you fewer sales than it saves in returns and bad reviews. A customer who reads that a product does not fit their model and does not buy is a review you never receive. Then reinforce it in the box Add a small insert card that repeats the solution. If the complaint is that setup is confusing, the card explains setup. If it is that people miss a feature, the card points at it. If it is sizing, the card explains care or fit. This catches the customer who did not read the listing carefully — which is most of them — at the exact moment they are forming the opinion that becomes a review. Include a route to reach you on the card, so a frustrated customer contacts you rather than going straight to the review box. That pairs with the support tactics in Short-Term Rating Recovery. Watch what happens Conversion rate may dip slightly at first. That can be the fix working: you are filtering out buyers who would have been disappointed. Fewer, better-matched customers is the intended outcome. Watch new reviews rather than the overall average, which moves too slowly to tell you anything for months. And watch returns, which respond faster than reviews and point the same way. See Understanding Business Reports and Sales Performance: Which Products Carry the Business. The wider use This exercise is worth running even when your rating is fine. Your competitors' negative reviews are the best differentiation research available — see Designing the Offer — and your own are the clearest brief for a main image — see Finding Reference Images. Reviews are the only place customers tell you, unprompted and at length, exactly what is wrong. Common mistakes Reacting to the loudest review instead of the most frequent complaint. Treating a quality complaint as a content problem. Better copy does not fix a product that breaks. Addressing it in one place, which most shoppers will not see. Vague reassurance rather than specific numbers. Skipping the insert card, which reaches the customer at the moment the review forms. Related - Fix the Product by Studying Competitors — for the complaints content cannot fix. - Short-Term Rating Recovery — holding the line while this takes effect. - Why a Rating Drop Costs More Than It Looks — the cascade this prevents.

Optimising Your Variation Order

On a listing with variations, the one shown first gets a disproportionate share of the clicks and the conversions. It is pre-selected, it is what the main image shows, and most shoppers never touch the swatches. Amazon sorts variations alphanumerically. So unless you have intervened, the order is an accident of naming — and your best seller may be sitting at the bottom. Watch the walkthrough Prefer to watch? One hundred and forty-four seconds on why the first variation gets the clicks, how to control the order, and choosing by conversion rather than by units sold. What this actually costs Say you sell baby towels in four designs, and White Puppy is comfortably your best converter. Alphabetically, the order runs Brown Puppy, Grey Puppy, Pink Puppy, White Puppy — last. Every organic shopper who lands on that listing sees Brown first. Your highest-converting option is three clicks away, and most people never make them. The product is fine. The listing is fine. The ordering is quietly taxing both. You can control the order Because the sort is alphanumeric, changing the variation name changes the position. Prefix the name with a number: 1. White Puppy, 2. Pink Puppy, and so on. Or use an allowed leading character — - White Puppy, ~ White Puppy — in categories that permit it. Numbers are the safer choice. They read as intentional to a shopper, they sort predictably, and they are accepted everywhere. Special characters are cosmetically cleaner but not universally allowed, so check your category before relying on them. How to do it 1. Open the listing's variations in Seller Central. 2. Identify the genuine best performer — by sales, click-through rate and conversion rate, not by intuition or by what you personally like. 3. Edit the variation name so it begins with a number or an allowed character. 4. Save, then refresh the live listing. 5. Confirm it now appears first. 6. Monitor conversion afterwards. Step 2 is the one people get wrong. The variation you think sells best and the variation that actually converts best are frequently not the same, and promoting the wrong one makes the listing worse. Picking the right variation Compare your variations against each other, not against the parent. Parent-level numbers average across children and hide exactly the differences you are trying to act on. Look at three things per child ASIN: units sold, click-through rate, and conversion rate. Prefer the best converter over the biggest seller when they differ. A variation that sells more only because it has been shown first is not evidence of anything — that is the bias you are trying to correct. Conversion rate is the cleaner signal. And check the rating too. Promoting a variation with a weak rating puts your worst reviews in front of every visitor — see Find the Variation Dragging Your Rating Down. Per-variation performance, side by side Pulling child-level sales, CTR, conversion and rating side by side is exactly what the product table is for — see Understanding the Product Details Insights Page. Doing the same comparison by exporting reports per child ASIN is an afternoon of assembly for a number you should be able to read at a glance. Then verify it worked Watch the parent's conversion rate for the following two to four weeks. Expect a modest, real improvement, not a step change. You have not changed the product or the traffic — you have changed which option the traffic meets first. If conversion falls, change it back. That is a legitimate outcome: it means the variation that was leading really was the right one, and your performance read was wrong. When to revisit it Seasonally, where variation demand shifts — colours and sizes rarely perform evenly across the year. After adding a new variation, which will otherwise slot in wherever the alphabet puts it. After a price change on one child, which moves its conversion rate relative to the others. Common mistakes Never having looked. By far the most common — the order is a default that no one chose. Promoting the biggest seller when it is only the biggest seller because it was already first. Reading parent-level data and missing the differences between children entirely. Promoting a variation with a poor rating, putting your weakest reviews in front of everyone. Using special characters in a category that rejects them, and having the change silently fail. Not checking the live listing afterwards. The edit is not the outcome; the displayed order is. Related - Find the Variation Dragging Your Rating Down — the rating side of the same comparison. - Understanding the Product Details Insights Page — where the per-child numbers live. - Understanding Business Reports — session and conversion data per ASIN.

Packaging Design for Amazon

Packaging is one of the most underrated conversion levers on Amazon, and it is the only one that works in five places at once. Higher CTR, because a listing with professional packaging stands out in search results. Higher conversion, because premium-looking products are trusted more. Better reviews, because a clean unboxing experience prompts them. Fewer returns, because good design communicates quality accurately and sets correct expectations — see Reducing Your Return Rate. Stronger brand recall, which drives repeat purchase. Watch the walkthrough Prefer to watch? One hundred and eighty-eight seconds on what belongs in a packaging brief, testing the design on live Amazon traffic, and updating the listing so the change actually reaches shoppers. Write the brief before you commission anything Nearly every disappointing packaging project is a brief problem, not a designer problem. Include all of this: Product overview — what it is and why it is different. Target audience — demographics, lifestyle, preferences. Pull this from Brand Analytics → Consumer Behavior Analytics → Demographics rather than assuming. Competitor references — screenshots, annotated with what you like and what you do not. Visual style — modern, clean, playful, earthy, clinical. Pick a direction and commit to it. Required elements, and be exhaustive: Logo. Barcode placement. Icons — recyclable, country of origin, certifications. Supplement or nutrition facts, if applicable. Product dimensions. Include around five reference designs, not twenty. A wide reference set produces averaged, characterless work. Giving feedback that produces a better draft Be specific and actionable. Not: "make it pop." Instead: "use a bolder serif for the product name", or "move the icons closer to the bottom edge". Rank the concepts, and eliminate the off-brief ones early so the effort concentrates where it is useful. Then shortlist three to five, iterate, and choose. What to collect at the end Print-ready dielines — AI and PDF. PNG and JPEG versions for Amazon use. Fonts, colour codes, and editable source files. That last one matters most. Without editable files, every future variation, size and marketplace version means starting again. Test it before you print thousands Consumer preference testing will tell you which of two designs people prefer, on a real sample rather than on your own taste. And Manage Your Experiments lets Brand Registry sellers A/B test on live Amazon traffic — put the new packaging in the main image and run it against the current one. That second test is the more valuable of the two, because it measures what actually happens to your CTR and conversion rather than what people say they prefer. See Testing a New Main Image. What is worth testing: Front panel layout. Colour scheme. Icon placement. Callouts — the specific claims on the front. CTR and conversion, before and after the design change A packaging change shows up as a CTR and conversion movement across your top keywords, and reading that against a baseline is what makes the test conclusive — see Reading Organic and Sponsored Rank. Align it with the listing, not just the product Put your primary keyword on the front panel. If you sell organic ashwagandha gummies, those words belong on the box — it is what the shopper is scanning for in the image. Then carry the same visual language across everything: Main images. Storefront header. Brand Story module. A+ content. Packaging that looks different from your listing is a wasted asset. The point is that a shopper recognises the same product across search, listing and delivery. After you pick a winner Re-shoot the main image with the new packaging prominent — or build a mockup — see Why the Main Image Decides Your CTR. Update lifestyle images so the packaging appears in use. Add insert cards with instructions, warranty information and a route to contact you — which is also how you intercept a return before it happens — see Short-Term Rating Recovery. Refresh A+ content and the storefront to match. The packaging is not the deliverable. The updated listing is. Common mistakes Commissioning from a thin brief, then blaming the output. Twenty reference designs, producing something averaged and generic. Vague feedback, which cannot be acted on. Not collecting editable source files. Printing before testing. Leaving the keyword off the front panel. New packaging with old listing images, so none of the conversion benefit reaches the shopper. Related - Why the Main Image Decides Your CTR — where packaging pays off first. - Testing a New Main Image — how to prove the change worked. - Reducing Your Return Rate — the expectations packaging sets.

Dealing With Listing Hijackers

Offers appearing under Other Sellers on Amazon on your own listing mean one of three things — and they need completely different responses. Diagnose before you act. Sending a cease-and-desist to Amazon Warehouse achieves nothing except wasting the week you needed. Watch the walkthrough Prefer to watch? One hundred and seventy-one seconds on telling the three cases apart, the four removal methods in order of strength, and verifying removal rather than trusting a closed case. Three things it could be 1. A hijacker selling counterfeits Poorly made knock-offs on your listing. They take your sales, attract bad reviews that attach permanently to your ASIN, and damage the brand. The clearest signal is in your reviews: complaints describing a product that is not the one you sell. Wrong material, wrong size, missing parts, poor build — from buyers who genuinely received something else. This is the one you attack with everything. 2. A legitimate reseller Selling your actual product, sourced somewhere real. Quality is intact, but they take your sales, and winning the Buy Box back usually means cutting price — which costs you margin on every unit, not just theirs. You cannot report them for counterfeiting, because they are not counterfeiting. Only serialisation keeps them off the listing. 3. Amazon Warehouse Amazon reselling sellable customer returns, in a condition other than New. Not a problem to solve. It is a normal consequence of returns — see Reducing Your Return Rate if the volume bothers you. What it costs while it runs You lose the Buy Box, so orders placed on your listing go to them. Negative reviews accumulate on your ASIN from customers who bought a counterfeit — see Why a Rating Drop Costs More Than It Looks. Brand image and repeat purchase suffer. And winning the Buy Box back means cutting price, which reduces profit on every unit you sell — not just the ones they took. Buy Box and rating, watched per product A hijacking shows up first as a Buy Box loss and a rating slide, often before you notice the offer itself — see Find the Variation Dragging Your Rating Down. Noticing it from a monthly review means a month of lost sales and permanent reviews. The four methods, in order of strength 1. Project Zero — the strongest Self-service removal. Enrolled brands remove a hijacker from their listing without needing Amazon's approval on each case. That is the whole difference. Every other method is a request that Amazon may or may not action, at a speed you do not control. Eligibility requirements are not especially demanding. If you are eligible, enrol — it changes hijacking from a multi-week case into a same-day action. 2. Cease and desist A message through Seller Central telling them to stop. It works more often than it should, because the recipient worries about account suspension and frequently just removes the offer. Free, immediate, and worth sending while slower methods run. 3. Test buy and IP violation report Buy the unit yourself, so you have evidence of what they are actually shipping. Then file through Report a Violation in Brand Registry, or the Report Infringement form. If Amazon confirms infringement, the seller is removed. The test buy is the part people skip and the part that makes the report credible. A report without evidence is an assertion. 4. Transparency Product serialisation. Every unit carries a unique code, and a product enrolled in Transparency cannot be listed or shipped on Amazon without a valid code. This is prevention rather than removal, and it is the only thing that stops legitimate resellers as well as counterfeiters. Slower to set up, because it changes your packaging and your supplier's process. Worth it on products that are repeatedly targeted. Which to use | Situation | Methods | |---|---| | Counterfeit hijacker | All four — Project Zero, cease and desist, IP report, Transparency | | Legitimate reseller | Project Zero, Transparency only | | Amazon Warehouse | None needed | Run them in parallel, not in sequence. Send the cease and desist today, file the IP report today, and let Project Zero remove them immediately if you have it. Verify the removal Open your product detail page and look below the Buy Box. If the New & Used offers section shows no other sellers, they are gone. Check it yourself rather than trusting the case status. A closed case is not the same as a removed offer — see Escalating an Amazon Case. And re-check periodically. Removed hijackers return, frequently under a new seller name. Common mistakes Acting before diagnosing, and treating a reseller or Amazon Warehouse as a counterfeiter. Filing an IP report without a test buy. Running the methods one at a time, when they can all run at once. Cutting price to win the Buy Box back, and leaving it cut after they are gone. Not enrolling in Project Zero while eligible, and handling every incident as a case. Checking once. They come back. Related - Why a Rating Drop Costs More Than It Looks — the review damage a hijacker leaves behind. - Escalating an Amazon Case — when the report stalls. - Short-Term Rating Recovery — repairing the rating afterwards.

Removing Reviews That Break the Rules

A meaningful share of the one- and two-star reviews dragging your rating down are not about your product at all. They are about delivery time, or the state of the box, or Amazon's handling — and those are not permitted in a product review under Amazon's guidelines. Which means they can be removed. Watch the walkthrough Prefer to watch? One hundred and sixty seconds on which reviews actually qualify for removal, why it takes repeated cases, and pricing a removal service against the product. What actually qualifies A review must be about the product — its features, design, quality, or the difference between what was described and what arrived. Reviews that are not about the product can be reported. Typical examples: "One star. Amazon said two days and it arrived after two weeks. I bought it for a birthday." "My product arrived in an Amazon box that was soaked, ripped and open." Neither says anything about the product. The first is a delivery complaint; the second is a fulfilment complaint. Both are legitimate frustrations aimed at the wrong target. This is the same principle as removing seller feedback that is really about Amazon — it is simply harder on product reviews, and takes more persistence. The reality of the process It is a case, and it is rarely resolved on the first one. Success comes from persistence: filing, following up, filing again, and eventually reaching a representative who actions it. Which is exactly why this is worth structuring — see Escalating an Amazon Case. Doing it yourself is free and slow. Services exist that do it at scale, using automated analysis to find which of your reviews violate the guidelines and file on your behalf. If you use a service, price it properly They typically charge per review removed — around $200–250 at last check, and that is negotiable. Whether it is worth it is arithmetic, not principle. Six one-star reviews removed in a month at $250 each is $1,500. On a product carrying real profit, that is straightforwardly positive — see Why a Rating Drop Costs More Than It Looks for what a tenth of a star is worth. Two rules if you engage one: Watch the bill. It grows quickly, because they are paid per success and have every reason to keep going. Scope which star ratings they work on. On a product earning $100k a month in profit, tell them to clear one, two and three stars and set a five-figure budget. On a product earning $5k a month, restrict them to one and two stars and accept a longer payback period. Set the scope before they start, not after the first invoice. Rating and its effect, per product Whether this is worth doing at all is a per-product question, and the input is what each product's rating is actually costing you — see Find the Variation Dragging Your Rating Down. Where this sits in the wider work Removal is a short-term lever, not a fix. It works because those reviews should not have been there — it does not address reviews that are about your product and are accurate. So run it alongside the real work: Short-term recovery — see Short-Term Rating Recovery. Content that answers the recurring complaints — see Answer the Complaints in Your Content. Product fixes where the complaint is genuine — see Fix the Product by Studying Competitors. Removing ten off-topic reviews from a product with a genuine defect buys you a few weeks. The defect keeps producing new ones. Common mistakes Reporting reviews that are genuinely about the product because they are negative. That is not what the guidelines cover, and it wastes the cases you need. Giving up after one case. Persistence is the mechanism. Engaging a service without a scope, and paying to remove three-star reviews on a low-margin product. Not watching the bill. Treating removal as the fix, and never addressing why the reviews exist. Related - Why a Rating Drop Costs More Than It Looks — what makes this worth paying for. - Short-Term Rating Recovery — the other fast levers. - Answer the Complaints in Your Content — the durable fix.

Intercepting a Return With Product Support

The moment a customer decides something is wrong, they are already on the Orders page — one click from a return, and two from a one-star review. Amazon's product support programme puts a "Get product support" button on that same page, so the customer can reach you before they reach either. Every conversation you have there is a return you did not fund and a review you did not receive. Watch the walkthrough Prefer to watch? One hundred and sixty-seven seconds on putting a third option beside Return and Review, which product profiles it pays off on, and turning chat logs into listing fixes. What it is Live chat and support, embedded on Amazon's own order page, through a provider partnered with Amazon's product lifecycle support programme. You need Brand Registry. And it is a paid service, so the arithmetic below decides whether it makes sense for your catalogue. What the customer sees On the Orders page, alongside Return and Review, a Get product support option. Choosing it lets them visit your support site or start a live chat with you, and in many setups submit a replacement claim instead of a return. That is the whole mechanism: the same decision point, with a third option on it. What it lets you do Fix a misunderstanding. A large share of returns and negative reviews come from a customer who could not work out how something worked, or expected something the product was never meant to do — see Reducing Your Return Rate for the three-way split of return causes. Send a replacement instead of taking a return. Cheaper for you, better for the customer, and it does not carry the return's rank and rating cost. Automate the common cases. Chatbots and canned flows handle the recurring questions — missing part, wrong size, setup step — without a person waiting. And it happens before the return, not after. Which is the entire value: by the time a return is filed, the fee, the fulfilment cost and the sentiment are already spent. Where the leverage actually is This is worth most on products with a specific profile: Anything with setup, assembly or a learning curve. Anything with sizing or compatibility questions. Anything where a missing component makes the whole unit useless — a screw, a cable, an adapter. High-value products, where the cost of one avoided return is large. It is worth least on simple products that either work or do not, where there is nothing to explain. Returns and rating, per product Deciding whether this pays means knowing your return rate and rating per product, not across the account — see Understanding Business Reports. One product with a confusing setup step can be carrying the whole account's return rate. Doing the arithmetic Take your worst product on returns. Estimate what a return costs you — the refund, the original fulfilment fee, the returns processing fee where you are over the category threshold, and the rank and rating damage. Then estimate what share of those returns are confusion rather than defect. Reading the return reasons and the negative reviews tells you. If a meaningful share are confusion, this pays. If nearly all are genuine defects, it does not — fix the product instead, see Fix the Product by Studying Competitors. It is not a substitute for the listing work Every question a customer asks through support is a question your listing failed to answer. So treat the chat log as research. The recurring questions belong in your images, your A+ content and a bullet — see Answer the Complaints in Your Content. Support intercepts the individual customer. The listing intercepts all of them. Run both: support for the ones already in trouble, listing changes so fewer arrive there. Common mistakes Enabling it and not staffing it. An unanswered chat is worse than no chat option. Using it on simple products where there is nothing to explain, and paying for nothing. Never reading the chat logs, and missing the clearest list of listing problems you will ever get. Treating it as a return-prevention tool only. The review it prevents is often worth more than the return. Skipping the arithmetic, and paying a monthly fee against a return problem that turns out to be defects. Related - Reducing Your Return Rate — the three causes, and which one this addresses. - Short-Term Rating Recovery — the other ways to intercept a bad review. - Answer the Complaints in Your Content — turning support conversations into listing fixes.

The Rating Recovery Action Plan

A rating falling from 4.3 to 4.2 is not a cosmetic problem. It starts a negative flywheel: conversion and click-through fall, sales fall, TACOS rises, organic rank slips, and each of those makes the next one worse. So the sequencing matters as much as the tactics. This is the order to run them in, and how hard to run them. Watch the walkthrough Prefer to watch? One hundred and sixty-six seconds on the two things to do on day one, sequencing content against the competitor teardown, and sizing the effort to what the drop is costing. Day 1, before anything else: remove low-rated child ASINs This is the only action that can restore your rating the same day. Removing one to three badly rated child variations from a parent routinely takes a listing straight back from 4.2 to 4.3 — and that buys you the weeks or months you need to fix the actual problem. Check every marketplace, not just your main one. If the listing has a single child, there is nothing to do here. With five, ten or fifty children, this is the first thing to look at — see Find the Variation Dragging Your Rating Down. And be clear about what it is: relief, not a fix. The underlying cause is still producing reviews. Day 1, second: order your competitors Order the top three well-rated competitors and have them shipped to your supplier. Do this on day one, because it takes one to two weeks to arrive and the clock starts when you order, not when you get to it. Order a set to yourself as well if you want to examine them directly — useful, not essential. This is the long-term fix in motion while you work on everything else — see Fix the Product by Studying Competitors. The next two weeks: content While the competitor units are in transit, analyse your negative reviews and find the top one to three recurring problems. Then answer them in the listing. An image for the biggest problem, an A+ section, a bullet, and the video where relevant. This takes real time — a good image or A+ section is not an afternoon — which is exactly why it fits the shipping window. Then implement it live — see Answer the Complaints in Your Content. And create insert cards addressing the same problems, before your next purchase order — because they have to be manufactured with the units. The next PO By the time you place it, three things should be true: The product quality issue is understood, from the competitor teardown. The improved content is live. The insert cards are designed and ready to include. That PO is where the fix actually lands. Everything before it is holding the line. Short-term levers, in priority order Run these alongside the long-term work: 1. Review-removal tooling for guideline violations — simplest to implement, real impact. 2. A removal service for off-topic negative reviews — costs money rather than time, which makes it easy to run in parallel — see Removing Reviews That Break the Rules. 3. Vine — slower, and worth it when the product genuinely matters — see Vine: Reviews for a New Product. 4. Product support on the order page — for large products only, given the cost — see Intercepting a Return With Product Support. The order is by effort-to-impact, so you are always doing the cheapest useful thing next. How hard to push depends entirely on the product This is the judgement call, and it is arithmetic. If this is your leading product, run everything, and make it the first thing you do each day — one to three hours — until the rating recovers. A product doing $5m a year that drops to 4.2 and loses 30% of its sales has cost you $1.5m. That justifies near-total attention. A product doing $500k inside a $5m brand is worth fixing, but not at that intensity. Run the long-term work, plus child-ASIN removal, plus the cheap review-removal levers, and let it take longer. If sales have fallen dramatically, move faster and use everything. If they have barely moved, you have time. What the rating drop is actually costing, per product The whole prioritisation depends on knowing what this specific product's rating drop is costing in sales, not on how the rating looks — see Why a Rating Drop Costs More Than It Looks. The goal Get back to 4.3. Not 4.5, not perfect — back over the threshold you fell below, because that is where the flywheel reverses. Common mistakes Starting with the long-term work and skipping the same-day child-ASIN check. Ordering competitor products late, so the teardown lands after the next PO. Treating child-ASIN removal as the fix, and stopping there. Designing insert cards after the PO is placed. Running every lever on a product that does not justify it, or too few on one that does. Not prioritising at all on a leading product, and letting the flywheel run for a quarter. Related - Why a Rating Drop Costs More Than It Looks — why this is urgent. - Find the Variation Dragging Your Rating Down — the same-day action. - Short-Term Rating Recovery — the fast levers in detail.