NEW

NEW

Adbrew MCP is now available for all customers.

Learn more here

Amazon Backend Keywords: 2026 Rules, Limits and Examples

Shrey Chaudhary

TL;DR

Key takeaways

  • Amazon backend keywords go in one Generic keyword field per ASIN, limited to less than 250 bytes (249 at most) in the US.

  • Spaces and punctuation don't count toward the limit. Brand names, ASINs, and subjective or temporary words are prohibited.

  • The best backend words are search terms that already convert in your PPC ads but aren't in your title, Item Highlights or bullets.

  • Since July 27, 2026, titles are capped at 75 characters, so words cut from titles need a new home.

Amazon backend keywords are hidden search terms you add to a listing so it can be found for searches its visible copy doesn't cover. In the US the field holds less than 250 bytes, which in plain English letters is up to 249 characters, spaces not counted.

The harder question is what to put in those bytes. This guide covers the 2026 rules, the July 2026 title change that made the field matter again, and how we fill it from search terms that already sell in our ads.

What Are Amazon Backend Keywords?

Amazon backend keywords (Amazon calls them search terms, entered in the Generic keyword field) are words attached to an ASIN that shoppers never see. Amazon's help page on search terms defines them as "keywords that are used by the search engine for lexical matching between an ASIN and a query."

Lexical matching means word matching. If a shopper types "kennel pad" and neither your title nor your bullets say "kennel", a backend "kennel" is what lets your listing be considered at all.

Backend keywords make a listing eligible for a search. They don't make it rank. Rank still comes from relevance, sales and conversion on that search, which is why the words you choose matter more than how many you fit.

The word "keyword" means three different things on Amazon, and mixing them up causes real mistakes. Our guide to Amazon keywords vs search terms separates them in full; the short version is below.

Backend keyword

Advertising keyword

Customer search term

What it is

Hidden listing word for organic search

Word you bid on in a campaign

What the shopper typed

Where it lives

Seller Central, Generic keyword field

Amazon Ads console

Search term report, Brand Analytics

Who chooses it

You

You

The shopper

Limit

Under 250 bytes per ASIN (US)

No fixed length per campaign

None

Costs money

No

Yes, per click

No

The Amazon Backend Keywords Limit: 249 Bytes

Amazon's guide to optimizing product discoverability says: "We limit the length of the Generic Keyword attribute to less than 250 bytes." Less than 250 means 249 bytes is the most you can use, and three details decide how much text that really is.

  • Bytes, not characters. English letters and numbers are one byte each. Amazon notes that characters such as German umlauts can take two or more bytes.

  • Spaces and punctuation don't count. Amazon's FAQ says search does not count spaces or punctuation toward the length, and asks you to separate words with spaces for readability.

  • The field stops you. The Generic keyword field in Seller Central stops accepting input when the byte limit is reached, so you can't save an over-limit entry there.

Amazon backend keyword byte math: letters and numbers are one byte, ö and ß two, the euro sign three, and spaces are not counted against the 249-byte limit

We still write to fit under 249 bytes counting spaces. If it fits with spaces, it fits without them, and the string survives being pasted into a flat file or another tool that counts differently.

Why Older Guides Say 500 Characters

They describe a field that no longer exists. Amazon reduced the search terms fields from five to one "to improve the quality of Amazon search results and reduce the amount of superfluous search terms," and combined the old content into the single field.

The previous version of this post said 500 characters. It was wrong, and we've corrected it. Treat any guide quoting 500 characters for the US as out of date.

The limit is set per marketplace, and the help page we quote applies to the US store. If you sell in Europe, Japan or India, open the same help page in that marketplace's Seller Central before you write.

Which Product Types Must Fill It

Since December 6, 2023, the Generic keyword field is required for 23 product types, including blankets, board games, balloons, bedding sets and body care products. For everything else it is optional, but leaving it empty wastes the only hidden space you get.

What Amazon Allows and Prohibits in Backend Search Terms

Amazon's rules are short, and the help page is clear that breaking the prohibited list "may result in your ASIN being suppressed and may put the health of your selling account at risk."

Amazon backend search term rules for 2026: what to use and what is prohibited, including brand names, ASINs, temporary and subjective words

Do

Don't

Synonyms ("earbuds" for headphones)

Your brand or any other brand name

Spelling variations ("grey" if your title says "gray")

ASINs

Abbreviations ("xl", "bpa")

Temporary words: "new", "on sale", "limited time", "latest"

All lowercase, separated by spaces

Subjective claims: "best", "cheapest", "amazing", "popular"

Singular or plural, not both

Profanity or offensive terms

Words not already in your title or bullets

Articles and short words: "a", "and", "for", "the", "with"

One rule contradicts a lot of advice still circulating. Amazon's US page says to use spelling variations "but don't include common misspellings." Many older guides tell you to fill the field with typos; Amazon's own guidance now says not to.

Amazon also says it "may not use the search terms that you supply for any reason," and that terms deemed relevant can change over time. Saving a word is no guarantee it is indexed, which is why we check (more on that below).

Other Backend Fields in 2026: Item Highlights, Item Type and Attributes

The Generic keyword field isn't the only place words go that aren't your title. In 2026 it sits in a wider system, and the biggest change came from the title rules.

1. Titles Are Now 75 Characters

On June 10, 2026, Amazon announced in Seller Central that from July 27, 2026, titles in all categories except media "will need to be 75 characters or less including spaces." Titles still over the limit are being replaced with Amazon's AI recommendation gradually.

Amazon's title requirements page now lists 75 characters as a requirement, alongside a limit of two uses of any word. A 200-character title stuffed with search terms is gone, and so are the words it carried.

2. Item Highlights: 125 Searchable Characters

The same announcement introduced Item Highlights: "an additional 125 characters for sharing materials or recommended use cases." Amazon says this content "is searchable and visible with titles in search results and on product detail pages."

Item Highlights is visible, so it isn't a backend field. It is, though, the first place the materials and use cases cut from your title should go. Our Amazon product titles guide covers how to write both.

3. Item Type Keyword

Item type keyword is a classification field, not a search terms field. Amazon's discoverability guide says it places your product in the most relevant subcategory. Get it right, but keep shopper keywords out of it.

4. Product-Type Attributes

Depending on the product type, the Product details tab can show attributes such as Target Audience, Intended Use or Subject Matter. Amazon's discoverability guidance doesn't describe these as keyword fields, and it doesn't publish how search weighs them.

We fill them with accurate, specific values ("senior dogs", "crate") and never with keyword lists. Wrong attributes can hide you from filters, and filters are increasingly how shoppers and AI answers narrow a search.

Where words go on an Amazon listing in 2026: 75-character title, 125-character Item Highlights, bullets, the hidden Generic keyword field and structured attributes

Here is the order we fill them in:

  1. The title gets brand, product type and the one or two terms with the most volume.

  2. Item Highlights gets materials, sizes and use cases cut from the title.

  3. Bullets and the description cover features and benefits in shopper language.

  4. The backend field gets everything relevant that is still missing: synonyms, variants, abbreviations and adjacent uses.

How to Fill Backend Keywords From Your PPC Search Terms

Your search term report is a list of real searches, with real orders, for your exact product. That is better evidence than any keyword tool's estimated volume.

If a converting search term's words aren't in your listing, you are paying for a search your listing could also win organically.

1. Pull 60 Days of Search Terms

In the Amazon Ads console, create a Sponsored Products search term report for the last 60 days, summary view. Amazon's report documentation sets a 65-day lookback and only includes terms with at least one click.

Pull Sponsored Brands too, and include auto and broad campaigns. Those are where new language shows up first. Our search term report guide has the download steps.

2. Keep the Terms That Convert

Filter to the same rule you'd use for harvesting: we use two or more orders at or under target ACOS (advertising cost of sales). A term that converts in ads is one Amazon already treats as relevant to your product.

Drop the terms that only spend. A term with 12 clicks and no orders doesn't belong in your listing any more than it belongs in your campaign.

3. Remove Rows That Can Never Go in the Backend

Three kinds of converting rows are banned from the field:

  • Competitor brand terms. They may convert as ads; they are prohibited as backend words.

  • Your own brand. Also prohibited, and your brand is in your title already.

  • ASIN rows. Product-page placements show as ASINs in the search term column. They belong in product targeting, not in your listing.

4. Split Terms Into Words and Remove What You Already Have

The field is a bag of words: Amazon asks for single words separated by spaces, with no repeats. So break each term into words. An n-gram analysis does this at scale and shows which single words carry the orders.

Then delete every word already in your title, Item Highlights or bullets. Amazon asks you to avoid repetition, and a repeated word spends bytes for nothing.

5. Add Variants and Synonyms, Then Order Them

Add the variants shoppers use that you don't: "grey" for "gray", "sofa" next to "couch", "xl" as well as "extra large". Write in lowercase with single spaces and no punctuation.

Put the words with the most orders first. If you ever need to trim, you trim from the end.

Worked example turning a dog bed's PPC search term report into a 111-byte backend keyword string

Here is the worked example from the image (example numbers). A dog bed titled "PawRest Orthopedic Dog Bed for Large Dogs, Washable Cover, Gray" has ten search terms with spend in the last 60 days, against a 30% target ACOS.

Customer search term

Orders

ACOS

Decision

orthopedic dog bed large

9

22%

All words already in title

dog crate mat

5

24%

Add "crate mat"

washable pet bed

4

27%

Add "pet"

senior dog bed arthritis

3

19%

Add "senior arthritis"

dog couch bed xl

3

25%

Add "couch xl"

snoozpaw dog bed

2

28%

Competitor brand: never in backend

b0c1xxxxxx

2

30%

ASIN: product targeting instead

best dog bed for big dogs

2

26%

Add "big", drop "best"

dog bed grey

2

21%

Add "grey" as a spelling variant

raised dog bed

0

n/a (12 clicks)

Not converting, not this product: skip

The PPC-sourced words plus a handful of synonyms produce this string: "crate mat pet couch xl senior arthritis big grey kennel pad lounger cushion mattress sofa joint hip older extra".

That is 19 words and 111 bytes with spaces, 93 without. It leaves more than half the field empty, and that is fine. We'd rather leave 130 bytes free than fill them with words no shopper uses for this product.

6. Fill Organic Gaps From Brand Analytics

Ads only show the searches you already bid on. If you are brand registered, Brand Analytics shows searches across all of Amazon, and the Search Query Performance dashboard shows the queries where your ASINs got impressions and clicks.

Look for queries where you get impressions but almost no clicks, and check whether their words are in your listing. Our guides to Brand Analytics and Search Query Performance walk through both reports.

7. Save, Wait, Then Check Indexing

Paste the string into the Generic keyword field (Manage All Inventory, Edit, Product details tab) and save. For many ASINs, use the Category Listing report and an inventory file with the partial update option, as Amazon's help page describes.

Give it a few days, then check whether each new word is indexed (method below). Re-run the whole routine every quarter, and after any title rewrite.

Loop between Amazon ads and backend keywords: discover search terms, keep converters, harvest into exact keywords and add missing words to the listing

Why Backend Keywords Matter for Your Ads

Backend keywords are an organic lever, but the reason we care is the ad bill. A search you win organically is one you don't have to buy on every visit.

Harvesting a converting term into an exact keyword buys that search; putting its words in your listing lets you also compete for it organically. Our keyword harvesting funnel guide covers the paid half.

Amazon's Sponsored Products targeting guide also says automatic targeting is "based on previous shopping queries and your product information." It doesn't say whether the hidden field counts as product information, so we don't rely on that.

Backend Keywords, Alexa for Shopping and AI Search

On May 13, 2026, Amazon brought Rufus and Alexa+ together as Alexa for Shopping. Shoppers can now ask questions in the main Amazon search bar and get "category and product insights in search results and on product pages."

So does a hidden list of words still matter when an AI assistant answers the shopper? Here is what Amazon has said, and what it hasn't.

What Amazon keyword search and Alexa for Shopping read, according to Amazon, and what Amazon has not published about the Generic keyword field
  • Keyword search is still lexical. The search terms help page, current in 2026, still describes the field as used "for lexical matching between an ASIN and a query."

  • The assistant reads your catalog content, reviews and Q&A. Amazon Science describes Rufus, the system behind Alexa for Shopping, pulling from "customer reviews, the product catalogue, and community questions and answers" before it answers.

  • Amazon models how products are used. Amazon's COSMO research builds a knowledge graph linking products to their functions, audiences and uses, such as slip-resistant shoes for pregnant women.

  • Amazon hasn't said whether the assistant reads the Generic keyword field, or how much weight it gives it. Anyone claiming a "Rufus backend strategy" is guessing.

Our approach follows from that. The backend field is your lexical safety net for words and variants. The visible copy, Item Highlights and accurate attributes are where you answer the questions an assistant asks about use, audience and fit.

Ads follow the same shift. Amazon Ads says existing sponsored ads campaigns are automatically eligible to serve in Alexa for Shopping, and Sponsored Products and Sponsored Brands prompts, generally available in the US since March 25, 2026, run on your existing targeting. Our Rufus guide covers the shopper side.

How to Check if a Backend Keyword Is Indexed

Amazon doesn't offer an official indexing checker. The test most practitioners use, including us, is a search on Amazon for your ASIN plus the word.

  1. Wait a few days after saving.

  2. Search Amazon for your ASIN followed by one backend word, for example "b0c1xxxxxx kennel".

  3. If your product appears, the word is very likely indexed for that ASIN. If nothing appears, it probably isn't.

Treat it as a strong signal, not proof. If a word that follows the rules never indexes, Amazon may have judged it not relevant; replace it.

The real test is commercial. In Search Query Performance, watch whether impressions start to appear on queries containing the new words over the next four to eight weeks.

Mistakes We See in Real Accounts

  • Copying your ad keyword list into the field. Your bid list repeats your title and includes brand terms. The backend wants what your listing is missing, nothing else.

  • Competitor brand names. They convert as ad targets and are prohibited as backend words. We find them in a lot of audits.

  • Filling the field with misspellings. Amazon's US guidance now says to skip common misspellings. Use real spelling variants instead.

  • Repeating title words. Every repeat wastes bytes and adds nothing to matching.

  • Commas, semicolons and capitals. They don't count toward the limit but don't help either. Lowercase words with single spaces are what Amazon asks for.

  • Setting it once in 2022. Your search term report changes every quarter, and the July 2026 title rules push words out of long titles. Re-check your backend after any title change.

  • Treating one ASIN's field as shared. Each child in a variation family has its own field. Use it for that child's size, color or use.

How We Run This in Adbrew

Everything above can be done by hand with a search term report and a spreadsheet. On one brand it's an afternoon a quarter. Across a portfolio, the backend review is the job nobody gets to.

Adbrew is an agentic retail media management platform for Amazon and Walmart, and the search term side of this routine runs inside it:

  • One view of search terms across every campaign. Filter converting search terms across Sponsored Products and Sponsored Brands in the Sponsored Ads management workspace, by product, without stitching reports together.

  • Harvesting on rules you set. The same converting terms you take to your listing are moved into exact match and negated at the source, on thresholds you control.

  • Agents for the reading. Ask Adbrew Intelligence for "search terms with two or more orders under 30% ACOS in the last 60 days for the dog bed ASINs, excluding brand terms", and export the list for your listing update. The same request works from Claude or ChatGPT through our MCP server.

Brands on Adbrew cut ACOS by 46% on average. Much of that comes from this kind of work: paying for the searches that sell, and making the listing win more of them on its own.

Let Your Ads Write Your Backend

The best backend keywords aren't found in a keyword tool. They are already in your search term report, converting at your target ACOS.

Want to see converting search terms across every campaign in one place, and harvesting running on rules you control? Book a demo and we'll walk through your search term report with you.

Frequently Asked Questions

What are Amazon backend keywords?

Amazon backend keywords are hidden search terms a seller adds to a listing in the Generic keyword field in Seller Central. Shoppers never see them, but Amazon search uses them for lexical matching between the ASIN and a query, so the listing can appear for words its title and bullets don't contain.

What is the Amazon backend keywords limit?

The Amazon backend keywords limit in the US is less than 250 bytes, so 249 bytes at most, in one Generic keyword field per ASIN. English letters and numbers count as one byte each, characters like ä count as two or more, and spaces and punctuation are not counted.

Do spaces count toward the 250-byte limit?

No. Amazon's search terms help page says spaces and punctuation are not counted toward the length of search terms. Amazon still asks sellers to separate words with spaces for readability. We write to fit under 249 bytes including spaces, so the entry stays valid in any tool or file that counts differently.

Where do I add backend keywords in Seller Central?

In Seller Central, open Manage All Inventory, choose the product, select Edit, then open the Product details tab and enter your words in the Generic keyword field. For many ASINs at once, download the Category Listing report and upload changes with an inventory file using the partial update option.

Should I put competitor brand names in backend keywords?

No. Amazon prohibits brand names in backend search terms, both your own and other brands, along with ASINs. Breaking these rules can get the ASIN suppressed and put account health at risk. If a competitor's brand converts for you, target it with Sponsored Products or product targeting instead.

Should I include misspellings in Amazon backend keywords?

Not common misspellings. Amazon's US search terms guidance says to use spelling variations but not to include common misspellings, and to use either singular or plural, not both. Real variants such as "grey" and "gray" or "xl" and "extra large" are the right use of the space.

How do I know if a backend keyword is indexed?

Amazon has no official checker. A common practitioner test is to search Amazon for your ASIN plus the word, for example "b0c1xxxxxx kennel". If your product appears, the word is very likely indexed. Then confirm commercially by watching for impressions on matching queries in Search Query Performance.

Do backend keywords still matter with Alexa for Shopping?

Yes, for keyword search. Amazon still describes backend search terms as used for lexical matching. Amazon says its shopping assistant draws on catalog content, reviews and Q&A, but it hasn't said whether it reads the Generic keyword field. Fill the backend for words and variants, and the visible listing for questions.

Take your Amazon PPC advertising to the next level

Book a demo

Related blogs

Grow your business on Amazon with Adbrew

Adbrew’s Amazon & Walmart PPC Optimization Platform is used by 5,000+ brands, sellers and agencies to optimize millions of dollars in ad spend every month.

1B+

Annual ad sales optimized

5000+

Brands & Agencies

46%

Avg. ACOS Decrease

Grow your business on Amazon with Adbrew

Adbrew’s Amazon & Walmart PPC Optimization Platform is used by 5,000+ brands, sellers and agencies to optimize millions of dollars in ad spend every month.

1B+

Annual ad sales optimized

5000+

Brands & Agencies

46%

Avg. ACOS Decrease

Grow your business on Amazon with Adbrew

Adbrew’s Amazon & Walmart PPC Optimization Platform is used by 5,000+ brands, sellers and agencies to optimize millions of dollars in ad spend every month.

1B+

Annual ad sales optimized

5000+

Brands & Agencies

46%

Avg. ACOS Decrease