What you have to be able to prove about a health claim
Read this before you write that a treatment works. It is the federal rulebook deciding whether a sentence on your site is a claim you must be able to back, what counts as backing, and what changed in 2024 for the reviews at the foot of the page.
- 1938 to 1994The statute takes its shapeSection 12 arrives in 1938 and makes false advertising of food, drugs and devices unlawful. In 1994 Congress adds one word, services, and a practice selling only its own time is inside it.
- July 26, 2023The Endorsement Guides are rewrittenThe version in force, pulling reviews, social posts and anything a reader takes to be somebody else's opinion into the FTC's reading of Section 5.
- October 21, 2024Reviews stop being guidanceThe Consumer Reviews and Testimonials Rule takes effect. Guides describe how the FTC reads the statute; a rule is a rule, and breaking one carries its own penalty.
Does it reach you?
Almost every practice owner assumes the FTC is for supplement sellers. Four situations, and three of them are you.
There is no list of formats
The FTC's health guidance says advertising covers not only television, radio, print and internet ads but “promotional materials such as brochures or booklets”, social media, “press releases, press interviews, or other media appearances”, and claims carried “indirectly through healthcare practitioners or other intermediaries”. If it exists to bring somebody in, it counts.
Second, the unit of judgement is the page, not the sentence. The FTC assesses the net impression left by all of it: headline, photographs, credentials, and what is missing. A page whose every sentence is true can still make a claim nobody wrote.
Seven lines decide almost everything
Four documents sit behind this page, and between them seven sentences do the work. Left, the text as written. Right, what each means when the advertisement is your own site.
Four words that carry the weight
The seven lines reuse the same terms, and most disagreements about a draft are about one of them.
Material
Would it change what a reasonable person thinks or does? Judged from the reader's side, which is why “we never meant that” is not an answer.
Substantiation
Evidence you hold before the claim goes out, about the claim you actually made, strong enough that the field would accept it. Nothing is filed, nothing approved.
Net impression
What the page conveys as a whole, photographs and layout included. The unit of judgement, and why a true sentence is not on its own a defence.
Clear and conspicuous
Defined in the Guides as difficult to miss and easily understandable by ordinary consumers, and on a web page, unavoidable.
“Competent and reliable scientific evidence”
Five words that stop some practices publishing at all, usually because they are read as an instruction to run a study.
We would have to run our own trial before we could write that.
Nobody expects a four-clinician practice to run a trial. The evidence can be somebody else's and nearly always is. What cannot be somebody else's is the holding of it: it has to exist, it has to be about the claim you made rather than a neighbouring one, and somebody has to have read it first.
Tests, analyses, research, or studies that have been conducted and evaluated in an objective manner by experts … and are generally accepted in the profession to yield accurate and reliable results.
The Commission's words, middle shortened. It adds that the research must be sufficient in quality and quantity against the whole body of relevant evidence, and that a health benefit claim will generally need randomised, controlled human testing.
The bar moves with the claim, not the page. “We offer this procedure” is checkable on its face. “This procedure relieves the condition” is a claim about what happens to patients. “Nine in ten of ours improve” is harder again, because the guidance says you must have at least the level of support you claim, and a number claims a study.
A disclaimer does not rescue a claim. The guidance is blunt: “Results not typical” does not cure the deception an unusual result creates, and a note at the foot of the page is not prominent enough to qualify it. What works is saying what a typical patient can expect, beside the claim and in the same size.
Your reviews are covered by a rule now, not a guide
Testimonials were always inside Section 5. Since October 2024 a separate rule governs how reviews are written, bought, solicited and suppressed, and the difference is not academic: breaking a rule carries a civil penalty for each violation, where deception alone does not until the Commission has already told you to stop.
Asking is fine; manufacturing is not. The rule says a generalised request to people who actually used the service sits outside the parts that would otherwise catch you, so inviting patients to review you is not the risk. The risk is a review that was never genuine, one bought for its sentiment, or a bad one made to disappear. And a testimonial is still a claim: a five-star review saying the pain went away asserts the treatment took the pain away, and the Guides put the practice behind that.
Replying is a different rulebook. Nothing here stops you answering a genuine review. What governs the wording of a reply is privacy: confirming in public that the person was your patient is the problem, not the reply. The privacy side →
There is no five year clock here, which is the harder version
Some regulators tell you how long to keep a copy of an advertisement. The FTC sets no such period for a practice's website, which reads as relief until you follow it through: the obligation is to hold the support while the claim is up, so a page nobody has opened in years is a claim you are making this morning.
Once the page is down the claim stops being made and what is left is the record. Whether you can answer for it depends on having kept the page as it stood and the evidence as it stood, not only the current version.
The part practices miss. A page written years ago and never revisited is being put out every day it stays up, so the question is not whether the evidence was good when somebody wrote it. It is whether it is good now.
What Verand checks of this, and what it does not
Everything above holds whether or not you buy anything. This is how much of it software can see in a draft, and the right hand column is longer, which is the reason for printing both.
What it checks
- Stops a guaranteed or permanent cure. A draft promising a guaranteed, permanent or complete cure, result or recovery cannot publish, nor can one saying a treatment cures cancer, diabetes, Alzheimer's, autism or arthritis. The one hard stop the medical rules add, and nothing releases it.
- Stops six more from the baseline every regulated site carries: guaranteed results, guaranteed returns, cannot lose, get rich quick, risk free, zero risk. The last two release on a written, logged reason from a named person. The rest do not.
- Flags the superlatives without stopping them. Best dentist, number one clinic, painless dentistry. Whether one is defensible turns on your state board, so a person settles it. Marking is all it does.
- Looks for the medical disclaimer and counts cited sources against a minimum of four. Both mark the draft rather than hold it.
What it does not
- It does not hold your evidence. Nothing here knows whether the study you have in mind supports the sentence beside it. Substantiation is a file at your practice.
- It does not check how old your support is. Nothing here knows whether what sits behind a claim has been superseded since you read it.
- It does not read for patient identity. The privacy scan looks for a Social Security number and a date of birth. Two patterns. Not names, not conditions, not a case somebody local would recognise, and we would rather print that than let you assume otherwise.
- It does not touch your reviews, your profile or your endorsers. Whether a review was bought, written by your hygienist or quietly taken down is not visible in a draft, so none of the 2024 rule is checked, and nor is disclosure.
- It does not publish. Nothing reaches your site without somebody at your practice approving it, and no setting changes that.
Validated against the FTC's rules. AI-researched and operator-reviewed. Your counsel confirms applicability. Not legal advice. How we label review →
The ones that come up
About the rules, not our software. Each points at the source it comes from.
Can we publish a patient's testimonial?
Yes, under two conditions that catch most practices out. It has to be the person's honest experience, and you have to hold the evidence for what it asserts, because a testimonial is read as a claim the treatment did what the patient says. If their result beats most, say what most get, beside it.
Is a before and after gallery a claim?
Yes. The statute counts representations made by design and device alongside those made in words, so photographs carry claims as sentences do. And a gallery of your best outcomes represents that those outcomes are what the treatment produces.
We are a non-profit clinic. Does any of this reach us?
Possibly not, and it is worth asking counsel rather than assuming either way. The Act limits the corporations the Commission reaches to bodies organised to carry on business for their own profit or that of their members. What does not move is your state's deception law, or your board's rule.
Can we ask happy patients to leave us a review?
Asking patients generally is fine, and the rule says so: a generalised request to people who used the service sits outside the parts that would otherwise catch you. It turns when you ask your own staff or their relatives, and even then the failure is the missing disclosure rather than the asking. What you cannot do is attach the request to a sentiment, because rewarding a positive review is a violation on its own.
Does it matter that a draft was written with AI?
Not to the FTC. Neither statute asks who typed it. The practice that puts the page up answers for what is on it, exactly as it would had an agency written the words. What changes is where the effort goes: producing the draft stops being the work, checking it becomes it.
Where every line came from
Primary sources only. Quoted phrases are the statute's or the regulation's own words, shortened where marked. Everything else is our plain English rendering.
Section references are to the source named beside them. Last checked against these sources on September 17, 2026. Not legal advice.
The next questions this one raises
Verand for medical and dental practices
What the software does, and refuses to do.
/medical-practices ExplainerWhen HIPAA calls your content marketing
The privacy side. A different regulator asking a different question.
/hipaa-marketing-rules ExplainerIs a medical blog “marketing” under HIPAA?
The scope test, on the privacy side of the same page.
/what-counts-as-advertising QuestionCan doctors, dentists and therapists use AI to write blog posts?
What the rules ask of a draft, whatever produced it.
/using-ai-to-write-content GuideLocal SEO for medical and dental practices
The pages about your office are advertising too.
/local-seo TrustHow we label compliance review
Operator-reviewed, not attorney-verified, and why we print it.
/compliance-standardsProving one claim is a morning. Proving two hundred pages is the job.
Verand drafts in your practice's voice and runs the checks above before anything reaches a reviewer. Somebody at your practice still presses publish.
Packs are AI-researched and operator-reviewed against the governing body's published rules. Not attorney-verified. Not legal advice.