Open Review Listings Changes in Seller Central (brand owners have 14 days before Amazon's AI-written titles go live), or Manage All Inventory, then Edit and View enhancements, to see the proposals. Approve the good ones, edit the weak ones, and replace the rest with your own title of 75 characters or fewer, spaces included.
Doing nothing is the risky choice. Over-limit titles are not chopped at character 75: Amazon's AI writes a replacement from your existing title and swaps it in gradually, and if nobody is watching the review queue, that version becomes the live title. An automated rewrite can also drop the brand or misread the product type.
Amazon says the cap covers every category except media, and it took effect on 27 July 2026. The old single title field is now two: an Item name of 75 characters and a searchable Item highlights field of 125 characters. Combined, that is the same 200 characters as before, but it is split into a short name and a longer supporting line. Menu names in Seller Central shift from time to time, so treat the paths below as a guide and confirm them in your own account.
What Amazon changed, and what it did not
Three things moved at once. The title limit dropped from 200 characters to 75 for every category except media. The notice does not spell out which categories count as media, so if you sell books, music, film or games, check the product title requirements page in Seller Central Help for your category. A new field called Item highlights appeared, holding up to 125 characters, which Amazon describes as room for materials and recommended uses. And from 10 August 2026, highlights display beneath the item name on both desktop and mobile, although the number of characters visible varies with screen size.
Amazon has said that search and discoverability are unchanged and that both fields are inputs for search, with neither prioritised over the other. That is worth testing rather than trusting blindly, but it does mean the old habit of cramming every keyword into one long title no longer has a home. The shape of a good title is now: brand, product type, the one or two attributes that separate this variant from its siblings, and nothing else.
Non-compliant listings stay active, editable and searchable while the rollout runs through 2026. That gives you time, but not unlimited time. The longer a listing sits over the limit, the more likely it is that Amazon's version replaces yours.

The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowFinding what Amazon's AI has proposed
There are two places to look, and a small seller should check both in the first week.
- Review Listings Changes. This is the queue for brand owners. Amazon's notice says brand owners have 14 days before the changes are implemented to review, modify and approve them. If the queue is empty or missing, ask whoever administers your Brand Registry account which role your login holds and whether it can see the tool, before assuming you have nothing pending.
- Manage All Inventory, then View enhancements. Select a listing, open its Edit menu and look for View enhancements. This shows what Amazon's AI currently suggests for that title. It works one listing at a time, which is fine for ten products and painful for four hundred.
Sellers without Brand Registry are in a grey area. Amazon's announcement and its follow-up answers describe the 14-day window for brand owners and say nothing about a window for anyone else. The safe reading is to assume you have no window and to fix your titles before Amazon does. Every seller can see the changes Amazon has already applied: Amazon's Engage with Amazon thread on the update points to View Change History in Manage All Inventory. Amazon says you can update your listings at any time in Manage All Inventory or with the List Your Products tool, and that applies whether a change is already live or only proposed.
A 20-minute weekly review that catches bad rewrites
Put a recurring slot in the calendar. Fourteen days sounds generous until a public holiday, a stock count and a supplier problem arrive in the same fortnight. Twenty minutes every Monday is enough for a catalogue of up to about a hundred listings, provided you do the steps in the same order each time.
- Open Review Listings Changes and sort by the date a proposal appeared, oldest first. The oldest one is the one closest to going live.
- Copy the original title, Amazon's proposed title and the listing's ASIN into a review sheet, one row per listing. Add columns for the character count of each (the formula is
=LEN(B2)), a verdict and your replacement title. - Score each proposal against the five-point rubric below. Anything that fails a point goes to the override list.
- Approve the ones that pass. Edit the ones that nearly pass. Write fresh titles for the rest.
- Log the date you acted, so that a listing never sits unreviewed because everyone assumed someone else had done it.
The five-point rubric
- Brand present and spelt your way. Amazon recommends keeping the brand name in the Item name. A rewrite that drops it, or that substitutes the registered company name from the brand attribute (a "Ltd" or "Inc" in the title), fails.
- Right product type. If the rewrite calls a base layer a "shirt" or a tin of oil a "sauce", the AI has misread your category data. Fail.
- The variating attribute is in the name. Colour, size, pack quantity or capacity, whichever separates this child listing from its siblings, belongs in the 75 characters. Amazon recommends this itself. If it was pushed out, fail.
- No claims you cannot support. A rewrite that says "waterproof" for a showerproof jacket, or "organic" for an uncertified product, is a compliance problem you now own, because your account published it.
- Reads like a person would search. Read it aloud. If it sounds like a list of fragments, a shopper will scroll past it.
A worked example: an online clothing store with 40 over-length titles
Say a small online clothing store sells outdoor wear under its own brand, shown in the examples as the seven-character placeholder [Brand]. Forty of its listings have titles between 100 and 150 characters, built up over the years to catch every search term. After the deadline, Amazon's AI has proposed replacements for all forty. The owner, who has about two hours free, sets up the review sheet and works through it.
| Version | Title | Characters |
|---|---|---|
| Original | [Brand] Men's Waterproof Hiking Jacket Lightweight Breathable Packable Rain Coat with Hood Windproof Outdoor Navy Large | 119 |
| Amazon's proposal (illustrative) | Waterproof Hiking Jacket for Men, Lightweight Rain Coat | 55 |
| Owner's override | [Brand] Men's Waterproof Hiking Jacket, Navy, Large | 51 |
| Item highlights | Lightweight breathable shell, taped seams, packs into its own pocket, windproof hood, rain coat for hiking and commuting | 120 |
The proposal is within the limit and reads naturally, so on first glance it looks fine. It fails the rubric on one point that matters: the brand is gone. It also drops colour and size, which is a problem because this listing is one child of a variation family. A shopper comparing six near-identical jackets in search results needs the variating attribute in the name to tell them apart.
The override keeps the brand, the product type and the two attributes, and uses 51 of 75 characters. Everything else the old title was carrying (lightweight, breathable, packable, windproof, hood) moves to the highlights line, which comes in at 120 of 125. Nothing the old title said has been thrown away; it has been sorted into the two fields.
Out of the forty proposals, say twenty-six fail on the brand, eight on the missing variating attribute, three on a wrong product type, and three pass untouched. That is an illustrative split, not a statistic. The point it makes is general: an AI rewrite is good at making a title shorter and much less reliable at knowing which words carry the weight for your buyers.
Let AI do the first pass, with your rubric as the instructions
Forty rows by hand is tolerable; four hundred is not. A chat assistant such as Claude or ChatGPT can draft the override and highlights for every row, and you then check the output rather than write from scratch. The prompt below applies the rubric explicitly. Paste in rows from your review sheet, five to ten at a time, so you can read what comes back.
You are rewriting Amazon product titles to fit a new rule: the Item name
must be 75 characters or fewer including spaces, and a separate Item
highlights field takes up to 125 characters.
For each row, return:
1. Item name: brand first, then product type, then only the attributes
that distinguish this variant (colour, size, pack quantity, capacity).
No promotional words, no claims not present in my attributes column.
2. Item highlights: the remaining useful detail (materials, uses,
features) as short comma-separated phrases.
3. Both character counts.
4. A one-line note if you had to drop something from the original.
Rules: use the brand exactly as written in the brand column. Never use
a company suffix such as Ltd or Inc. If the attributes column does not
support a claim in the original title, leave it out and flag it.
Rows:
[paste: ASIN | brand | original title | attributes]
Here is the kind of output you might get back for one row. The figures are illustrative, and the counts are worth recomputing, because language models miscount characters routinely.
ASIN B0XXXXXXX1
Item name: [Brand] Merino Base Layer Top, Women's, Charcoal, Medium (56)
Item highlights: Merino wool, 200gsm, long sleeve, odour resistant, for hiking, skiing and layering (82)
Note: original said "anti-bacterial"; not in attributes, left out.
Two habits make this safe. First, never trust the counts the assistant reports: paste the results into your sheet and let =LEN() measure them. Second, read the note lines. The flag about "anti-bacterial" is the assistant doing exactly what you asked, and it is the sort of claim that causes a listing to be challenged. If you want a wider look at generating bullets and search terms the same way, the Amazon listing copy tutorial covers that side.
Splitting a title between the name and the highlights
The hardest judgement is what to keep in the 75 characters. A simple order of priority works across most categories, and it is worth writing down so that anyone on your team makes the same call.
| Priority | Goes in the Item name | Goes in Item highlights |
|---|---|---|
| 1 | Brand, exactly as registered | Nothing to repeat |
| 2 | Plain product type a shopper would type | Alternative names for the same thing |
| 3 | The attribute that separates this variant | Other attributes that apply to the whole family |
| 4 | Pack quantity or capacity if it affects price | Materials, finish, certifications you can prove |
| 5 | Nothing else | Recommended uses and comparison terms |
A specialist food importer selling cooking oil in catering sizes shows how this plays out. The old title ran to 106 characters: brand, "Extra Virgin Olive Oil Cold Pressed Large Catering Tin 5 Litres Bulk Kitchen Restaurant Value Pack". The new name is "[Brand] Extra Virgin Olive Oil, 5 Litre Tin, Cold Pressed", at 57 characters. The highlights line then reads "Cold pressed, first harvest, bulk tin for kitchens and delis, pours cleanly with the screw spout", at 96. The words "bulk", "kitchen" and "restaurant" have not vanished; they now work as comparison and use terms in the field designed for them.
A rule of thumb for variation families: write the parent title first, so the shared words are settled, then add only the variating attribute to each child. Doing it in that order stops forty children from each inventing their own phrasing for the same jacket.
Overriding in bulk with the Category Listings report
When the catalogue is beyond what you can open one listing at a time, Amazon's own route is a spreadsheet. The Category Listings report is pre-populated with your current data and, since the change, shows Item highlights next to Item name. The loop looks like this.
- Download the Category Listings report for the category you are working on.
- Add a
=LEN()column beside the item name column and filter for anything above 75. That is your to-do list, and it will include listings Amazon has not yet got to. - Feed the filtered rows to your AI assistant with the prompt above, five to ten at a time.
- Paste the approved names and highlights back into the file, re-check the lengths with
=LEN(), and save it in the format the upload tool expects. - Upload through List Your Products and open the processing report to see which rows were rejected.
A quick sum helps you plan the bulk pass. Reading an AI draft, fixing it and checking its length takes about a minute and a half per row once you are in a rhythm. Four hundred listings at 90 seconds each is 36,000 seconds, or ten hours of checking. Split that into five two-hour sessions, oldest proposals first, and the whole catalogue is done inside a fortnight without anyone working late. If your queue holds proposals that are already ten days old, do those rows by hand first and leave the AI pass for listings Amazon has not reached yet.
Amazon says listings are typically updated within eight hours of an upload. Treat that as a minimum wait, not a promise. Compare the live title on the product page with your spreadsheet the next day, for a sample of ten rows.
Watch for one known snag. Sellers replying in Amazon's own forum thread reported error code 100476, "This attribute 'Item Highlight' is currently unsupported", even after shortening a title. Amazon has not posted a fix there. One approach worth trying is to upload the shortened names on their own first, confirm they went through, and then upload highlights as a separate pass for the affected category. If it persists, raise it with Seller Support, quoting the code and the file name.
When a rewrite goes live and the returns start
Take an illustrative kitchen fitter who sells worktop trims and fixings through Amazon and learns the cost of an unwatched queue. The team assumed the owner was checking it; the owner assumed the marketing assistant was. Nine of their listings changed on the same day, and three of them lost the pack quantity. The pack-of-ten hinge covers and their single-cover sibling both ended up called plain "Hinge Covers". Buyers who wanted ten picked the cheaper listing, received one cover, and returns followed in the same week.
The way it showed up was in the returns reasons and a few messages asking where the other nine were, not in a Seller Central alert. The fix was simple: restore the quantity in the title, check the listing against the product page, and note the owner of the queue in writing. The lesson is that pack quantity and capacity are the attributes an AI most often treats as filler, and the attributes that most directly cause refunds when they vanish.
Build the safeguard into the process rather than relying on attention. A named owner, a Monday slot and a deputy for holidays cost nothing. For a small team, the single most useful column in the review sheet is "date checked".
Keeping claims factual when the title gets shorter
Shortening a title tempts people to swap facts for adjectives: "Best", "Premium", "Top Quality". Those words use up characters a shopper cannot act on, and before you approve a batch it is worth reading Amazon's current product title guidance in Seller Central Help for any words it restricts. Keep the name factual and let the highlights carry the useful descriptive detail.
The same caution applies to AI-written highlights. A model asked to make a jacket sound appealing will happily say "fully waterproof" when your spec says "water resistant to 5,000 mm". Your attributes column is the only source it should draw on, and your rubric's fourth point exists for exactly this reason. When in doubt, take the claim out, because you can add it back after confirming it with the supplier's data sheet.
If your account sells in several marketplaces, the same product may have different limits elsewhere. The multichannel listing tutorial shows how to keep one master record and generate each channel's version from it, which saves you rewriting the same jacket four ways.
Protecting your search position while you shorten
Amazon says discoverability is unchanged, but your own numbers are the better judge. Before you change a batch of titles, note for each listing its sessions, conversion rate and the position of two or three important search terms. After two weeks, compare. If sessions on a listing fall by a large margin and nothing else changed (price, stock, a competitor's promotion), the title is a suspect.
Change a handful of listings at a time, not the whole catalogue in one go, so that you can tell which change moved which number. Start with listings that already have steady traffic, because small percentage changes show up clearly there, and leave your slowest sellers for later. Customer language matters as well. If your reviews keep describing the jacket as a "cagoule" and your title says "rain coat", that word belongs in highlights, and the review analysis tutorial shows how to pull such terms out of the review text.
How to check the whole job worked
Finish with a short audit rather than a feeling that it is done.
- Length. Re-download the Category Listings report and filter for names above 75. The count should be zero, apart from any media items.
- Review queue. Review Listings Changes shows nothing waiting past about ten days.
- Spot check. Open ten live product pages on a phone. Is the brand visible, and does the name distinguish the variant? Is the highlights line showing, and does it read sensibly when cut short, as it may be on a small screen?
- Returns and messages. For the next month, scan return reasons for "not as described", which is the signature of a lost attribute.
- Ownership. One named person, one calendar slot, one sheet.
If you would rather set this up with someone who has done it for a catalogue like yours, that is the kind of job my AI implementation consultation covers.
Questions sellers ask about the 75-character rewrites
Does Amazon cut titles off at 75 characters?
No. Amazon's notice says titles still over the limit are updated gradually to its AI recommendation, which it writes from your existing title, rather than being cut at character 75. Listings stay active and searchable while this happens, so the risk is a wrong replacement going live, not a broken listing.
Do sellers without Brand Registry get a review window?
Amazon's posts describe the 14-day window for brand owners only and say nothing about other sellers, so assume you do not. Any seller can still edit a title at any time in Manage All Inventory or by uploading a corrected file through List Your Products, which replaces whatever Amazon proposed.
Can I put my keywords in Item highlights instead?
Yes, that is its purpose. Amazon describes the 125-character field as room for materials and recommended uses, and says it feeds search just as the name does. Keep the brand and the main product words in the 75-character name, and use highlights for the secondary detail.
Does this apply to books, music and video?
Amazon exempts media from the 75-character limit, but its announcement does not list which categories count as media. Books, music and video are the usual reading; confirm yours in the product title requirements page in Seller Central Help. Every other category is covered, so check any listing outside media even if its title was fine under the old rule.
How long do bulk title edits take to appear?
Amazon says listings are typically updated within eight hours of an upload. Allow a full day before you judge, and check the Seller Central processing report for rejected rows rather than assuming silence means success.
Further reads
- AI for Amazon Listing Images: Infographics Inside Amazon's Rules — Images and infographics that work inside Amazon's rules.
- How Small Amazon Sellers Use AI for Listings and Reviews — A wider look at where AI helps a small Amazon seller.
- Translate an Amazon Listing to German With AI: Keywords and Checks — Translate a listing without breaking its length limits.
- Does Amazon Allow AI-Generated Product Images? What Passes Review — The image rules that sit next to the title rules.
- How to Write eBay Listings With AI That Sell Faster — Titles and descriptions for a marketplace with its own limits.
- Etsy's AI Title Suggestions: Should You Accept Them? — The same accept-or-override question on another marketplace.
- AI Product Title Generator: How to Test One Before You Trust It — A side-by-side test of an AI product title generator on 20 of your own listings, with accuracy and keyword checks before any go live.
- Amazon A+ Content With AI: The Built-In Generator and Checks — How to use the AI Ready modules in A+ Content Manager, and a pre-submission review for claims, image text and facts.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Amazon Seller Forums posts 'Updates to improve your product titles begin on July 27' (updated 10 June 2026), 'Answers to your product title update questions' and the Engage with Amazon Q&A thread on the title update, all read on 5 November 2026, plus seller replies in the same threads for error 100476; check Seller Central for the current wording of the tools named here.