Amazon Restricted Keywords for Supplements: The Words That Trigger Suppression

Amazon does not read your supplement listing the way a customer does. Automated systems tokenize every field you control and compare the language against restricted-keyword lists, claim patterns, and category rules. One flagged word can suppress an ASIN in minutes, with no human involved until you appeal. Knowing which words trip the wire, and why, is the difference between a listing that sells and a listing that sits in search suppressed status while you guess at the cause.

This guide covers the categories of language that trigger enforcement for dietary supplements, what to write instead, and the trap most sellers fall into after their first suppression: swapping words without changing the claim.

Where Amazon actually looks

Sellers routinely sanitize their bullets and still get flagged, because detection is not limited to bullets. Amazon scans:

• Title, bullets, and description. The obvious surfaces, checked continuously, not just at listing creation.
• Backend search terms. Invisible to shoppers, fully visible to Amazon. Stuffing "diabetes support" or "flu remedy" into backend terms is one of the most common causes of suppression on otherwise clean listings.
• A+ content. Both the text modules and the text inside A+ images.
• Images. Text in infographics, badges, and rendered labels is read via OCR. An old infographic that says "fights inflammation" is a violation even if the bullets were fixed last year.

Amazon's Misleading and prohibited claims policy is explicit that products cannot claim to cure, mitigate, treat, or prevent disease, and enforcement applies across the entire detail page. Detection also catches close variants: misspellings, plurals, and hyphenated forms of restricted terms are flagged the same way.

The six categories of trigger language

1. Disease names

Arthritis, diabetes, cancer, COVID, anxiety, depression, ADHD, herpes, flu, insomnia as a condition. Naming a disease anywhere on the page implies your product affects that disease, which makes it an unapproved drug claim in FDA's eyes and a prohibited claim in Amazon's. This includes the sneaky forms: "not intended to treat arthritis" still contains the word arthritis, and a review-bait question like "does this help with diabetes?" in your A+ FAQ is still a disease reference you published.

Safer framing: describe the normal body structure or function instead. Joint comfort and mobility, healthy blood sugar levels already within normal range, occasional stress, restful sleep. FDA's structure/function claims framework exists precisely for this. Our guide to structure/function claims versus disease claims maps the line in detail.

2. Treatment verbs

Cures, treats, heals, prevents, remedies, reverses, fights, relieves. These verbs convert an innocent noun into a disease claim. "Supports joints" is a structure/function claim. "Relieves joint pain" is a treatment claim, because pain relief is a drug function.

Safer framing: supports, maintains, promotes, helps sustain. These are the standard structure/function verbs, and they work as long as the object of the sentence is a normal function, not a symptom or disease.

3. Drug comparisons

"Natural Xanax," "like Adderall," "herbal Viagra," "plant-based Ozempic." FDA treats a claim that a product substitutes for a drug as an implied disease claim, full stop. Amazon flags drug names outright, and the brand owners of those drugs file trademark complaints on top of the policy violation.

Safer framing: there is none. Delete the comparison. No rewrite of "natural Xanax" survives review, because the entire message is "this treats what that drug treats."

4. FDA-adjacent phrases

"FDA approved" is the classic. Dietary supplements are never FDA approved, so the phrase is false on its face and flagged instantly. "FDA certified," "FDA endorsed," and badge graphics with FDA logos fall in the same bucket. Even "FDA registered facility," which is factually accurate for compliant manufacturers, gets flagged when presented as a product endorsement.

Safer framing: if you reference the FDA at all, keep it to the required DSHEA disclaimer text. Save the facility registration language for your documents, where it belongs.

5. Pesticide trigger words

Antibacterial, antiviral, antifungal, antimicrobial, kills germs, sanitizes. This category surprises supplement sellers most. Under FIFRA, Amazon's pesticide marking policy treats any product marketed with pesticidal claims as a pesticide, regardless of what it is. An oregano oil listing that says "natural antifungal" gets pulled into pesticide review, where Amazon demands an EPA registration number that no supplement can ever have. The listing stays blocked until the language is gone.

Safer framing: drop the anti- words entirely and return to structure/function territory: supports immune function, supports the body's natural defenses.

6. Superlatives with proof burdens

"#1 doctor recommended," "clinically proven," "guaranteed results," "best on Amazon." These are not banned words so much as claims that require evidence you probably do not have. The FTC's Health Products Compliance Guidance requires that a "clinically proven" claim be backed by clinical proof, generally randomized controlled human trials on your product or an equivalent formulation. Amazon flags unsubstantiated superlatives, and the FTC fines for them. Our breakdown of what clinically proven actually requires covers the substantiation standard.

Safer framing: claim what your evidence supports. "Formulated with clinically studied KSM-66 ashwagandha" is defensible if the branded ingredient has trials at your dose. "Clinically proven results" is not.

Quick reference: risky term to safer framing

Risky term	Why it triggers	Safer framing
"Fights arthritis"	Disease name plus treatment verb	"Supports joint comfort and mobility"
"Lowers blood sugar"	Disease treatment (diabetes implied)	"Supports healthy blood sugar levels already in normal range"
"Cures insomnia"	Disease claim	"Promotes restful sleep"
"Natural Xanax"	Drug substitution claim, trademark	Remove entirely; "supports a calm mood"
"Anti-anxiety formula"	Names a condition	"Helps manage occasional stress"
"FDA approved supplement"	False; supplements are never approved	Remove; use the standard DSHEA disclaimer only
"Antiviral immune booster"	FIFRA pesticide trigger	"Supports immune function"
"Kills harmful bacteria"	FIFRA pesticide trigger	"Supports a balanced gut microbiome"
"Clinically proven to melt fat"	Unsubstantiated proof claim plus disease-adjacent	"Supports metabolism as part of a healthy diet"
"#1 doctor recommended"	Superlative requiring survey evidence	Remove unless you own the survey data

Why keyword swaps alone do not save you

Here is the mistake that gets sellers flagged twice. They run a find-and-replace, turn "treats anxiety" into "supports calm during anxious moments," and consider it fixed. It is not, for two reasons.

First, the condition is still named. "Anxious moments" reads as anxiety to both Amazon's classifiers and FDA reviewers.

Second, and more fundamentally, regulators judge the net impression of the whole listing, not a word list. The FTC's guidance is explicit that claims are evaluated in the context of the entire ad, and FDA evaluates labeling the same way: product name, images, testimonials, and surrounding copy all count. A product called RheumEase with a knee X-ray infographic and reviews praising arthritis relief is making a disease claim even if every sentence uses the word "supports." Automated detection catches words. The human who reviews your appeal catches meaning. You need to survive both.

This is also why suppression rarely travels alone. A keyword flag invites human review, and human review reads everything, including the ingredient claims Amazon is enforcing under its March 31, 2026 misleading claims deadline.

Find the triggers before Amazon does

If a listing is already suppressed, start with our guide to fixing a suppressed supplement listing, because the fix depends on which system flagged you. For everything still live, the job is to read every field, including backend terms and image text, against all six categories above. That is tedious to do by hand and exactly what VERID8 automates: it scans your listing text and images against FDA, FTC, and Amazon rules, flags each risky term with the reason, and writes a compliant replacement that keeps the selling point without the claim. You can check one listing free in about a minute.

The words on this page are not a list to memorize. They are a pattern: name a disease, promise a drug's job, or borrow authority you do not have, and the system will find it. Sell the function, own the proof, and the keywords take care of themselves.