Unapproved Health Claims: What Meta Actually Flags

What counts as an unapproved health claim on Meta?

On Meta, an unapproved health claim is language that turns a consumer product into a medical promise. The label on the jar matters less than the sentence in the ad. A capsule can be framed as a routine item, a beauty item, or a treatment claim; only the last one is dangerous. That is why the Meta ad review guide matters: it says the review looks beyond the headline and into the full ad experience.

I am inferring the exact classifier logic from Meta's published review surface. Meta does not publish the model rules. It does publish enough to show that the decision is not made on product name alone.

That is the point.

Why does the same product pass in one ad and fail in another?

The same product passes in one ad and fails in another because the claim stack changed. A magnesium ad that says 'nightly support' can clear review, while the same bottle with 'fix insomnia in 3 nights' can fail. The product did not change. The promise did. Meta's system can also revisit an ad after launch, so a creative that slipped through on day 1 can still get pulled later if the page or the edit changes.

That is why you have to think in pairs: ad plus landing page. If the ad is soft and the page is hard, the harder language can sink the whole submission.

No shortcut.

Which words and structures reliably trigger the classifier?

Words and structures do the damage. Meta's policy on drugs and pharmaceuticals says ads cannot explicitly or implicitly claim a product can treat, cure, prevent, mitigate, or diagnose a disease or medical condition. In practice, that means the risky part is not just the noun. It is the verb, the timeframe, the symptom list, and the before-and-after frame. Meta's restricted-content policy page gives the clearest public wording.

The classifier is not a dictionary. Context matters. But certain patterns raise risk fast.

If you have to ask whether a sentence sounds medical, it probably does.

How do timelines and quantified outcomes change the verdict?

Timelines and quantified outcomes change the verdict because they convert a vague promise into a measurable one. 'Supports a healthy routine' is broad. 'Drop 2 inches in 14 days' is a promise with a clock on it. Numbers are not the problem by themselves. Dose counts, serving sizes, and bottle counts are fine. Outcome numbers are the issue.

The FTC says health-related claims need solid proof, and its health products guidance warns marketers to review both the literal and implied message. The FTC health claims page and the FTC health products guidance make the same basic point from the regulator side. Keep the claim narrow or drop it.

That is the trap.

What is the difference between a health claim and a wellness statement?

A health claim says a product affects a body state, symptom, condition, or diagnosis. A wellness statement stays at the level of routine, preference, ingredient, or general support. 'Magnesium glycinate for a calmer bedtime routine' sits closer to wellness than 'magnesium fixes insomnia'. 'A daily capsule with zinc and vitamin C' is a description. 'Prevents colds' is a claim. The line is not the word 'natural'. The line is the implied medical effect.

Support words are not a shield. If the rest of the copy points to disease, the ad still lands in claim territory.

How does the policy apply to the landing page, not just the ad?

Meta applies the policy to the landing page, not just the ad. That is where a lot of nutra ads die. The review guide says the system can inspect the destination page, and it can review ads again after launch. So if your headline is clean but the page says 'rebuild cartilage', 'reverse prediabetes', or 'lose weight without diet or exercise', the ad can still fail. The click does not hide the page.

The Meta Ad Library is useful here, but only for what it is. It shows currently active ads across Meta products, and for political or issue ads it also preserves inactive ones. For regulated-niche research, that means you can inspect live angles, country variants, and whether a page is actively buying. It is not a complete archive of every angle ever tested. That gap matters.

• Use it to confirm a page is currently live.
• Use it to see which creative frames are public right now.
• Use it to compare country-specific versions.
• Do not use it as proof that a hidden variant is safe.

The page matters.

How do you self-audit a creative before submitting it?

The blandest copy is not the safest copy. A vague headline can still fail if the landing page carries the hard claim, while a tighter, specific line can pass because it stays away from diagnosis and outcome language. Safety comes from the whole claim stack, not from washing the sentence until it means nothing.

Use a manual audit before you submit. It takes 3 minutes, and most affiliates still skip it.

• Read the headline, primary text, and call to action out loud.
• Circle every disease, symptom, diagnosis, cure, prevent, treat, reverse, heal, and fast-result word.
• Search the landing page for the same words, plus the FAQ, testimonials, and footer.
• Check any image text, subtitles, and OCR overlays.
• Remove before-and-after panels unless you can defend every implied outcome.
• Keep one product, one promise, one page.
• Use the Ad Library to see current live phrasing from the category, not to copy it.

This manual method is boring. It works.

What Meta usually rejects is not the brand. It is the claim stack. If you tighten the sentence, match the page, and remove the medical promise, you reduce the odds of a rejection without guessing at the classifier.

FAQ

What usually triggers Meta's unapproved health claims review? Medical language does. If the ad promises diagnosis, cure, prevention, symptom relief, or a measurable body change on a deadline, Meta is more likely to flag it. The product can be ordinary. The sentence is usually the problem.

Can a supplement ad pass if it uses the word support? Yes, sometimes. 'Support' is not a magic shield, though. If the same creative or landing page also says treat, reverse, or lose weight fast, the claim can still fail because the surrounding message reads as a health promise.

Does the landing page matter if the ad copy is clean? Yes. Meta's review process can inspect the destination page and can review ads again after launch. If the page says something stronger than the ad, the stronger claim can sink the whole submission.

Is the Ad Library enough to research competitors? No. It is good for current live ads and public variants, and for political or issue ads it holds inactive ones too. It is not a full archive of every test a competitor ran, and it will not show private or deleted paths.

What is the fastest self-audit before upload? Read the ad and page side by side, then strike every disease word, timeline, and before-and-after cue. If a line sounds like a doctor would say it, remove it. That takes minutes and catches most avoidable rejections.