The gap between what sellers fear and what actually happens
Ask ten sellers why a listing got suppressed and most will say "keyword stuffing". In practice, suppression and search-visibility loss cluster around a much narrower set of causes, and nearly all of them are about claims, not keywords.
A claim is any statement a customer could reasonably act on: a performance number, a comparison, a health outcome, a certification, a guarantee. Amazon's policies do not prohibit strong copy. They prohibit claims you cannot substantiate, plus a specific list of things that are never allowed regardless of substantiation.
The four categories that cause real problems
1. Ranking and superlative claims
"#1 best seller", "top rated", "voted best" — these are prohibited in title and bullet copy unless they refer to a badge Amazon itself has awarded, and even then the badge is displayed by Amazon, not written by you. The reason is simple: rank is volatile. A claim that is true on Tuesday is false on Wednesday, and Amazon does not want to arbitrate that.
Instead of: "The #1 rated stainless steel kettle on Amazon" Write: "1.7L stainless steel kettle with a 2200W concealed element"
The second version is duller and converts better, because it answers the question the shopper actually has.
2. Health, medical and safety claims
This is the category that produces account-level action rather than a listing takedown, so it deserves disproportionate care. Words like cures, treats, prevents, heals, anti-bacterial, FDA approved, and clinically proven move a general merchandise listing into a regulated category.
"FDA approved" is the most commonly misused phrase in ecommerce copy. The FDA approves drugs and certain Class III devices. It does not approve cosmetics, supplements, food contact materials, or the overwhelming majority of consumer goods. A facility can be FDA registered, which means something quite different and is not a quality endorsement.
3. Fabricated social proof
"Over 10,000 happy customers", "as seen on TV", "recommended by dentists". If you cannot produce the underlying evidence within the hour, do not write it. Review-manipulation policy covers not just soliciting reviews but referencing review volume or sentiment in the listing itself.
4. Contact diversion
Any URL, email address, phone number, QR code, or instruction to "contact us directly for a better price" in listing content is a policy violation. It is also one of the easiest violations for automated systems to detect, because it is a regex.
Guarantees and warranties
You may state a warranty you actually offer. You may not state a guarantee Amazon would have to honour, and you may not imply Amazon endorses it. "30-day money-back guarantee" is fine if you operate one and can honour it under Amazon's returns framework. "Lifetime guarantee" almost always is not, because "lifetime" is undefined and unenforceable.
A rewriting method that works
Take an existing listing and mark every sentence with one of three labels:
- Fact — a specification you can evidence from a spec sheet, test report, or the physical product in front of you.
- Inference — something reasonable but unproven ("great for small kitchens").
- Claim — anything a regulator or Amazon would want substantiated.
Facts stay. Inferences stay if they are clearly subjective and non-comparative. Claims must either be evidenced in writing or deleted. Most listings lose 20-30% of their word count and gain conversion, because the deleted material was noise.
What to do if you are already suppressed
Fix the copy first, then appeal. Appeals that arrive before the violating text has been removed are almost always rejected, and each rejection lengthens the review queue. Keep the plan of action short: what the violating text was, why it violated the policy, what it now says, and what process change prevents recurrence.
This article summarises publicly documented marketplace policy as a starting point for your own review. It is not legal advice, and policy text changes — always check the current version in Seller Central before you rely on it.
Written by the CopyForge AI team. Drafted with AI assistance, then reviewed and fact-checked by a human before publishing.