Where exactly is the line between a cosmetic claim and a drug claim?
It is not a word list. The line turns on intended use - what your copy says the product is for.
- Cosmetic territory: cleansing, beautifying, altering or improving appearance. How skin looks and feels.
- Drug territory: anything implying the product treats, prevents, mitigates or cures a condition, or changes how the body works.
The implication is the trigger. That is why searching for banned words is the wrong instinct twice over: a sentence containing nothing from any list can still read as a drug claim, and the reverse. Two brands can sell the same formula and only one of them has a problem, because only one of them wrote the copy as a treatment.
Practically, this means claims are a writing discipline owned by whoever writes the listing - not a compliance step somebody runs at the end with a spreadsheet of words.
Does your listing copy count as evidence?
Yes, and this is the single most useful thing to know here. FDA has said intended use can be established by claims appearing in labeling, in advertising, on the Internet, or in other promotional materials. Your Amazon detail page is promotional material you published about your own product.
So the common assumption - the label was reviewed, therefore we are covered - does not hold. Three things follow:
- A compliant carton plus a non-compliant bullet is still a problem. The listing is not downstream of the label. It is a second set of claims with your name on it.
- Listing copy is usually written last, by a different person. The pack goes through review. The bullets get written by a marketer or an agency months later, often against a keyword brief, and are never sent back through the same sign-off.
- Old claims outlive the pack. A phrase you took off the carton two revisions ago can still be sitting in bullet four, in an A+ module, and in a Sponsored Brands headline.
The fix is procedural, not clever: whoever signs off on the packaging signs off on the detail page, including image text, and the sign-off is dated.
How do you rewrite a claim without losing the sale?
Appearance and feel survive. Treatment does not. The same jar can be described either way, and the two versions are not treated the same.
- Say what the customer sees, not what the product does to a condition. Reduces the appearance of fine lines rather than treats wrinkles. Skin looks clearer rather than clears breakouts.
- Watch the verbs that describe a body process. Anything positioned as acting on inflammation, bacteria, hormones or healing is describing how the body works, whatever the surrounding sentence says.
- Before and after framing is where copy drifts. A visual or a caption that reads as clinical improvement of a condition makes the claim, even when the text underneath is careful.
- Comparative language borrows the risk. Positioning your product against a treatment - as an alternative to, works like - imports the treatment claim into your listing.
The objection is always that the careful version converts worse. Usually the opposite is true, because escalating the verb is what copy does when it has nothing specific to say. Texture, absorbency, who it is for, how it layers under makeup, fragrance free, non comedogenic - that is what a beauty buyer is actually reading for, and none of it costs you anything in compliance.
The skill is maximising conversion inside the appearance envelope rather than testing the edge of it. Copy that lives on the edge has to be defended every time the enforcement bar moves.
Which of your SKUs are actually drug products?
Some of them genuinely are, in the US, no matter how the listing is written. Sunscreen and anything carrying an SPF, acne products with an active, anti-dandruff. Those are not copy problems you write around - they are a different build.
- Decide before you list. Rebuilding a listing after a flag costs you the rank and the velocity you built getting there. Determining status first costs you a week.
- Documentation is the gate, not the wording. These SKUs carry requirements a moisturizer does not, and Amazon asks for paperwork rather than argument.
- A formula change can move the SKU across the line. Adding SPF to a day cream is the obvious one. The listing was compliant for the previous formula.
- The rest of the catalog does not inherit the answer. Status is per product, so one approved SKU does not clear the sibling that shares its brand and its imagery.
Where this lands: it is a regulatory question, and it belongs with your regulatory advisor or counsel before it reaches your copywriter. What we do is make sure the listing plan reflects the answer instead of assuming one.
Where do claims live besides the bullets?
Everywhere you publish about the product. This is where otherwise careful brands get caught, because the audit stops at the text fields.
- Title, bullets and description. The part everybody checks.
- Text baked into images. Infographics, benefit callouts, chart-style comparisons in the image stack. Nobody can search a JPEG, so these are rarely re-read after upload - and a compliant bullet under a non-compliant infographic is still a non-compliant listing.
- A+ modules and brand story. Same problem, plus these are often built once by a designer working from an old brief. See A+ content for how these modules are built.
- Storefront pages and video. Published by you, about your product, on Amazon.
- Ad creative. Sponsored Brands headlines and custom images carry claims and are reviewed on their own track, so a headline can be the thing that draws attention to the page.
- Backend search terms. They are copy too. You wrote them, and they state what you intend the product to be found for.
Run the check as a sweep across every surface for one ASIN at a time, then across its variation siblings. Checking the hero SKU only is how a family gets taken down one child at a time.
Why is this a recurring check rather than a launch task?
Because the target moves and you do not get told. Amazon does not publish its restricted-term list and it changes without notice, so copy that cleared review last year can be flagged this year with nothing on your side having changed. The public Amazon guidance for sellers is blunt about the underlying rule: sellers may not sell products claiming to treat, cure or alleviate disease without FDA approval.
What a takedown costs is almost all downstream of the takedown itself:
- Rank and velocity decay while you appeal. You are not paused, you are falling, and the recovery is a second project after reinstatement.
- Review flow stops. The reviews that would have landed in that window do not arrive late. They do not arrive.
- Ads keep buying clicks to a dead page until somebody notices and pauses them that day.
- Variation siblings inherit the damage, so a single flagged child can drag a family.
- Reinstatement runs on the Amazon timeline, not yours, which is the part no plan can compress.
So: a standing quarterly sweep, a re-check after any A+ refresh or new ad creative, and a dated change log of what the copy said and when - which is also the record an appeal needs. Write to the policy rather than around it, and argue about one bullet now instead of writing an appeal later. If something is already down, start with the suppressed listing fix.
Where to go next
Questions people actually ask
Have your claims read before Amazon reads them.
Read-only access, three days, a real report with real numbers — no obligation. If we can't find enough to justify our fee, we'll be the ones to say so.
No spam. No 12-field forms. Just a real reply from our team — usually the same day.
Prefer to talk first? Book a 30-minute call