Draft listings from a fact sheet so they fit Amazon's 2026 limits, a 75-character title plus up to 125 characters of Item Highlights in most categories, using Amazon's own tools or a chat assistant. Then use AI to find what reviews praise and criticise, fix the product or listing, and get more reviews only through Request a Review or Vine.
Two 2026 changes shape how this works. Amazon cut titles to 75 characters in every category except media from 27 July, and said it would gradually replace longer titles with AI-recommended ones. And Amazon's shopping assistant, renamed from Rufus to Alexa for Shopping in May 2026 where it's available, answers shoppers' questions using listing content, Q&A and reviews. Keyword-stuffed listings lose on both counts. Accurate, complete ones win.
What changed for Amazon listings in 2026
Amazon announced the title change in Seller Central in June 2026. The main points for a small seller:
- Titles: 75 characters, including spaces, in all categories except media (books, music and video). Amazon's stated reason is that titles then display in full on mobile.
- Item Highlights: up to 125 more characters for details such as materials or recommended uses. The content is searchable and shown with the title in search results and on the product page.
- AI recommendations. Amazon offers suggested titles and Item Highlights that fit the limits, and titles still over 75 characters after 27 July are being replaced with its AI recommendation gradually. Brand owners get 14 days to review suggested changes.
- The 2025 title rules still apply: no word used more than twice (articles, prepositions and conjunctions excepted), and no characters such as ! $ ? _ { } ^ unless they're part of the brand name.
Rules can differ between Amazon stores and categories, so check your category's style guide in Seller Central before rewriting a whole catalogue.
Start from a fact sheet, not from a competitor's listing
AI listings go wrong when the assistant fills gaps with claims that sound right. Give it a fact sheet and tell it to use nothing else. Here's one filled in for an illustrative small seller, a dog groomer who sells its own shampoo:
Product: oatmeal dog shampoo, 500ml bottle
Brand: [Brand]
Key ingredients: colloidal oatmeal, aloe vera
Free from: soap, added fragrance
Use: dogs and puppies over 12 weeks; dilute 1:4; lather, rinse
Who makes it: formulated and used in our own grooming salon
Tested: used on client dogs in our salon for two years
NOT claims we can make: vet-approved, hypoallergenic, treats itching
or any skin condition, organic
Packaging: recyclable bottle, flip-top capThe "not claims" line matters as much as the facts. Assistants reach for "vet-approved" and "hypoallergenic" because competitors use them, and health claims about treating conditions are ones you'd need to be able to back up. If it's not on the sheet, it doesn't go in the listing.
The sheet looks different for other products. A bakery selling boxed brownie-mix kits needs the allergen list copied exactly from the printed label, the shelf life, what the buyer must add (eggs, butter), tin size and yield. Its not-claims list might read "healthy, low sugar, homemade taste guaranteed". Food listings are where an assistant's helpful additions are riskiest, because an allergen or ingredient that doesn't match the label is a safety problem, not just a return. Paste the label text and tell the assistant to copy it, not summarise it.
Writing the title and Item Highlights with AI
Using ONLY the fact sheet, write:
- 3 Amazon titles, each 75 characters or fewer including spaces.
Brand first, then product type and the main benefit or use.
No word more than twice, no ! $ ? symbols, no promotional words.
- 2 versions of Item Highlights, each 125 characters or fewer, for
details that didn't fit in the title.
Give the character count for every line.
[paste fact sheet]An illustrative response and what's wrong with it:
Titles
1. [Brand] Oatmeal Dog Shampoo for Sensitive Skin, Unscented, 500ml (64)
2. Oatmeal Dog Shampoo, Sensitive Skin Dog Shampoo, Vet-Approved,
Unscented Dog Wash 500ml (87)
Item Highlights
A. Colloidal oatmeal and aloe, soap-free, fragrance-free, for dogs and
puppies over 12 weeks, made by a professional groomer (121)
B. Made by a professional dog groomer. Colloidal oatmeal and aloe,
soap-free, pH balanced for dogs. Suitable for puppies over 12 weeks. (132)Title 1 and Highlights A are usable. Title 2 breaks three rules at once: it's 87 characters, it uses "Dog" three times, and "vet-approved" is on the not-claims list. Highlights B is over 125 characters and adds "pH balanced for dogs", which isn't on the sheet. That pattern is typical: ask for several options, keep the one that obeys the sheet, and check character counts yourself, because assistants miscount.
Existing listings are the harder job, because a long title usually holds details someone once thought were essential. The salon's paw balm title before the change (illustrative) ran to 139 characters: "[Brand] Paw Balm for Dogs, Natural Dog Paw Balm with Shea Butter and Beeswax, Dog Paw Protector for Cracked Paws, Snow, Hot Pavements, 60ml". It uses "Paw" four times and "Dog" three times, so it failed the old repetition rule as well as the new limit. The rewrite keeps brand, product type, the two main ingredients and size: "[Brand] Dog Paw Balm with Shea Butter and Beeswax, 60ml Tin" at 59 characters. The uses (snow, hot pavements) move to Item Highlights, and "for cracked paws" is dropped entirely, because on reflection it reads as a treatment claim. Ask the assistant to list every detail it removed and where it went, so nothing useful disappears by accident.
If you sell sizes or scents as variations, check each child listing on its own. Each child listing carries its own title, whether you or Amazon's recommendation wrote it, and a size that falls out of one child's title leaves two options in the drop-down that look identical in search.
Bullets that answer the questions shoppers ask
Amazon's shopping assistant answers questions such as "is this safe for puppies?" or "does it smell?" from your listing, your Q&A and your reviews. If the answer isn't in your listing, it may come from a review instead, and reviews aren't always right. Write bullets that answer the questions buyers actually ask. You'll find them in your Q&A, buyer messages and reviews.
A before and after for one bullet:
Before: PREMIUM QUALITY - Our amazing shampoo leaves your dog's coat
soft, shiny and smelling great!
After: FRAGRANCE-FREE AND SOAP-FREE: made with colloidal oatmeal and
aloe, with no added scent, for dogs and puppies over 12 weeks. Dilute
1 part shampoo to 4 parts water; one 500ml bottle gives about 2.5
litres of wash.The first version answers nothing and contradicts the product ("smelling great" for an unscented shampoo). The second answers three common questions: does it smell, is it safe for my puppy, how long will a bottle last. Getting products recommended in ChatGPT shopping makes the same point for AI shopping assistants outside Amazon: clear, factual product information is what they draw on.
Buyer questions that arrive by message are a good test of whether the sheet is doing its job. A shopper asked whether the shampoo is safe for a cat. The assistant's first draft replied, "Yes, our gentle formula is suitable for cats too." Nothing on the sheet supports that; the product was only ever used on dogs. The reply that went out: "It's formulated for and tested on dogs, so we can't recommend it for cats. A shampoo made for cats would be the safer choice." The same question appeared twice more that month, so "for dogs only" joined the second bullet. A question asked three times is a gap in the listing, not a customer service task.
Once the listing is updated, check what the shopping assistant says about it. Where Alexa for Shopping is available, ask it on your own product page the questions buyers ask most, and compare each answer with the fact sheet. In one illustrative check, "can I use it on a puppy?" drew an answer quoting a review from someone who'd used it on an eight-week-old puppy, while the sheet says over 12 weeks. The fix was to make the age limit the opening words of the first bullet, then ask again a few weeks later. The assistant draws on whatever it finds, so the listing needs to say the right thing clearly enough to outweigh one enthusiastic reviewer.
Amazon's own AI tools or a chat assistant?
| Tool | What it does | Watch out for |
|---|---|---|
| AI in Add Products | Drafts titles, bullets and descriptions from images, a short description, a web page or a spreadsheet | Amazon says you're responsible for making sure it's accurate |
| Enhance My Listing | Suggests improvements to existing titles, bullets, attributes and descriptions; you accept, edit or decline | Suggestions can add claims your fact sheet rules out |
| A+ Content Manager AI | Generates text and images for A+ modules on brand-owned products | Needs Brand Registry; generated lifestyle images must match the real product |
| ChatGPT or Claude | Drafts from your fact sheet with your rules, in your voice, across many products | You must paste category rules and check limits yourself |
A practical split: let Amazon's tools suggest attributes and flag gaps, and write titles, highlights and bullets from your fact sheet with a chat assistant, so the not-claims list is always applied. Going through suggestions is quick once you sort them into three piles. For the detangling spray, an illustrative batch of Enhance My Listing suggestions included filling the missing "item volume" attribute (accept: it's a fact, and it feeds filters), changing "for tangled coats" to "for matted and tangled coats" (edit: the spray loosens light mats, so the salon kept "light mats" rather than a broader promise), and adding "vet-recommended formula" to a bullet (decline: it's on the not-claims list). Attributes are usually safe to accept after a glance; wording that changes what the product claims to do needs reading against the sheet every time. If you also sell on eBay, writing eBay listings with AI uses the same fact-sheet habit with eBay's 80-character titles.
Reading reviews with AI: yours and your competitors'
Reviews are the cheapest product research you'll get. Amazon already shows shoppers an AI summary of reviews ("Customers say") built from verified-purchase reviews, so the themes in your reviews are visible to every buyer. Know what they say before shoppers do.
Once a month, export or copy your recent reviews, and the reviews of two or three competing products, and ask:
Below are 120 recent reviews for my dog shampoo and 200 for two
competing products. For each product:
1. List the top five praise themes and top five complaint themes,
with how many reviews mention each and one short quote each.
2. Separate complaints about the product from complaints about
delivery or packaging.
3. Tell me which complaints about competitors my product avoids, and
which of my complaints they seem to avoid.
Don't infer themes from fewer than three reviews.An illustrative extract of the output:
YOUR PRODUCT - complaints
1. Cap leaks in transit (14 reviews): "arrived with shampoo all over the box"
2. Thin, runny texture (9): "feels watery, hard to lather"
3. Smell (4): "smells of oats, not nice" <- based on 4 reviews
COMPETITOR A - complaints
1. Strong perfume (22): "my dog smelled like air freshener for days"Two things to check. "Thin, runny texture" may be the dilution instruction confusing people rather than a product flaw, so read the actual reviews before changing the formula. And the leaking-cap theme is the biggest and cheapest fix of all: a better cap or a seal. Amazon's Voice of the Customer dashboard in Seller Central shows your offers' customer-experience health and the reasons behind negative experiences, so compare its return comments with what the assistant found. Spotting patterns in complaints, returns and defects with AI goes deeper into this analysis, and turning customer reviews into marketing copy covers using the praise themes in your bullets, in your own words.
Getting more reviews without breaking the rules
Amazon is strict here, and AI makes it easy to break the rules at scale. Based on Amazon's own guidance to sellers:
| Allowed | Not allowed |
|---|---|
| Using the Request a Review button in Seller Central, which sends an Amazon-generated message | Offering anything in exchange for a review: discounts, free products, refunds |
| Enrolling eligible products in Amazon Vine if you're brand-registered | Asking buyers to change or remove a review |
| Improving the product and listing so fewer buyers are disappointed | Seeking positive reviews through third-party services, social media groups or review sites |
| Answering buyer questions accurately and promptly | Writing, editing or posting reviews of your own or competitors' products |
| Contacting customers outside Amazon's approved channels to ask for reviews |
Vine sends products to Amazon's invited reviewers, who post honest reviews labelled as Vine reviews. It carries an enrolment fee that depends on how many units you enrol; check the current fee and eligibility in Seller Central before planning a launch around it.
A realistic mistake: a seller asks an assistant for "a friendly package insert that encourages reviews" and gets "Love it? Leave us a 5-star review! Not happy? Email us first and we'll make it right." That insert asks for positive reviews only and steers unhappy buyers away from reviewing, which is the kind of selective solicitation Amazon prohibits. What review gating is and why it gets businesses penalised explains the principle. If you use an insert at all, keep it neutral and tell the assistant the rule in the prompt. A neutral version might read: "Thank you for choosing our oatmeal shampoo. Dilute 1 part to 4 parts water for the best lather. If you have a question about using it, you can message us from your Amazon order page." It gives buyers something useful, asks nothing about ratings and treats every buyer the same. Check it against Amazon's current communication guidelines before printing a few thousand.
The first quarter for a groomer's shampoo line
Say the grooming salon sells three products on Amazon: the oatmeal shampoo, a detangling spray and a paw balm, about 180 orders a month in total.
Month one: listings. All three titles are over 75 characters. The owner writes a fact sheet for each (about 30 minutes per product), generates titles, Item Highlights and bullets with the prompt, and checks them against the sheets. Two hours in all, compared with letting Amazon's AI rewrite the titles, one of which had added "vet-recommended".
Month two: reviews. The monthly review prompt surfaces the leaking cap on the shampoo (14 of 120 reviews). The salon switches to a screw cap with an inner seal, at a few cents more per bottle, and adds "sealed cap" to the Item Highlights. The sum is quick: if the shampoo sells about 90 bottles a month and the new cap costs $0.08 more, that's $7.20 a month. Refunding or replacing even one leaked $15 order a week costs around $60 a month, before counting the one-star reviews. The texture complaints turn out to be buyers using it undiluted, so the dilution line moves to the first bullet.
Month three: results. Leak complaints stop appearing in new reviews, and returns for the shampoo fall. The owner uses Request a Review on every eligible order, which is quicker once it's part of the weekly routine. Total AI cost: a $20-a-month assistant. The biggest improvement came from a cap, which is typical: AI's most valuable output was finding the fix, not writing the copy.
Mistakes that get listings suppressed or accounts warned
- Claims the product can't support, especially health or safety claims added by an assistant. Keep a not-claims list and apply it to every prompt.
- Accepting AI suggestions in bulk without reading them, whether from Enhance My Listing or your own assistant.
- Keyword repetition and banned characters in titles. Count characters and repeated words before saving.
- AI-generated images that don't match the product. Lifestyle images from A+ tools or elsewhere must show what the buyer will actually receive.
- Any review scheme that isn't Request a Review or Vine. Including "friendly" inserts that ask only happy buyers.
Amazon listing and review questions
What happens if I don't shorten my Amazon titles to 75 characters?
Amazon said in its June 2026 announcement that titles still over the limit after 27 July would gradually be replaced with its AI-recommended version, while listings stay active. Brand owners get 14 days to review suggested changes before they go live. Writing your own compliant title and Item Highlights keeps the wording in your hands rather than the model's.
Does Amazon penalise listings written with AI?
Amazon offers its own AI tools for creating and improving listings, so AI drafting isn't the issue. What Amazon holds you responsible for is accuracy and compliance: every claim, attribute and image must be correct and follow its policies. Listings get into trouble for wrong or unsupported claims and rule breaches, whoever or whatever wrote them.
Is it all right to analyse competitors' Amazon reviews with AI?
Reading public reviews to understand what buyers like and dislike about similar products is ordinary market research, and AI makes it much quicker. Use the themes to improve your own product and listing. Don't copy competitors' listing text, don't make comparative claims about their products, and never post or arrange reviews about them.
Further reads
- How to Sync Stock Across Shopify, Amazon and eBay Automatically — Keeping stock in step if you sell on Amazon and elsewhere.
- AI Product Photos vs a Photographer: Cost and Quality Compared — Main images and lifestyle shots: AI or a photographer.
- How to Handle a Fake or Unfair Review With AI's Help — What to do about a review you believe is fake or unfair.
- How to Track Competitors' Prices and Offers With AI — Watching competitors' prices and offers over time.
- How to Write Shopify Product Descriptions With AI That Convert — The same product copy adapted for your own online shop.
- How to Use AI for Etsy Titles, Tags and Descriptions — Search phrases first, then prompts for each field: a title under 15 words, 13 non-repeating tags of 20 characters or fewer, and a description buyers can act on.
- Etsy's AI Title Suggestions: Should You Accept Them? — Etsy's title tool reads your title, first photo and description but not your sales. A four-question test for each suggestion, with worked verdicts.
- How Resellers Use AI to List Second-Hand Items Faster — A second-hand listing workflow that uses marketplace AI for the drafting and keeps measurements, flaws and photos honest.
- Does Writing Etsy Listings With AI Hurt Your Search Ranking? — How Etsy search actually ranks listings, the specific ways AI-drafted copy loses ground, and a simple test to run on your own shop.
- AI Review Management Tools for Small Businesses (2026) — Which review tools are worth paying for: verified prices, what their AI actually does, and picks for a garage, an online supplier and a multi-branch agency.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Amazon Seller Central announcement on product titles and Item Highlights (June 2026); Sell on Amazon guidance on AI listing tools; About Amazon on Enhance My Listing, AI review highlights and Rufus; Amazon Seller Forums staff post on product review policy; GeekWire on Alexa for Shopping, checked September 2026.