Insights

Health copy that promises nothing, and still says plenty

Health copy that promises nothing, and still says plenty

On November 5, 2026, I read every service and condition page on a psychiatry practice's site with one question in mind: which sentences would make a certification reviewer stop? The site had been written carefully. I still found 54 places worth a second look, and the fixes went live, in English and Russian, before the evening was over.

Why the read happened

The practice had decided to apply for LegitScript certification. For many kinds of health advertising, the big ad platforms will not run a practice's ads without it. PPC.land reported on September 27, 2026 that more than 10,000 websites now hold it, and that Google treats it as an entry condition for pharmacy and telemedicine advertising in eleven markets. Certified sites stay under monitoring afterwards, and that monitoring looks at licensure and at what the practice says online.

The site was already live and doing its job, so nothing was urgent. But an application fee is paid before the review starts and is not refunded if the answer is no. Reading the site first, the way the reviewer would, costs an evening. Being turned down costs the fee and a wait.

The standard behind all of this is older than any certifier. When the FTC finalized its order against a children's supplement brand on July 15, 2026, the core requirement was that any health or efficacy claim be "supported by competent and reliable scientific evidence." The same order dealt with reviews written by the company's own people and five-star reviews paid for with discounts. A psychiatry practice is a long way from a height supplement, but the rule it answers to is the same one.

What the pages that were already right had in common

Some pages needed nothing, and they are the best guide to the rest. The therapy page on internal family systems names its research, a small pilot with no control group and a randomized trial published in January 2026, and then says, in the practice's own voice, "we would rather tell you that than oversell it." The complex PTSD page says outright that nothing on it is a promise. The Spravato page sticks to the label's own indications and the safety program the drug requires, and carries the line "Results cannot be guaranteed."

None of those pages is timid. They explain a lot. What they do not do is turn research into a guarantee. That is a writing skill more than a legal one.

Where careful copy still promises

The 54 findings were four high, nineteen medium and the rest low, and most of them were small. They fell into a handful of patterns that I now look for on every health site.

Diagram: 54 findings from a claims read on November 5, 2026, grouped into patterns: a medicine described without its plain status, a saving offered for paying ahead, a rule described as good through a date eight weeks out, pages hinting at office visits before the office opens, early research written as settled fact, unmeasured comparisons with other clinics, and consented patient words that sound like results.
Seven patterns, none of them careless. Each one is a sentence that says slightly more than anyone has confirmed.
  • A mechanism stated as fact. "Research links this action to neuroplasticity" is true as far as it goes. "Research in animals and early human studies links this action to neuroplasticity" is true and tells the reader how far it goes.
  • A benefit list. A paragraph of things some patients describe reads, on a quick pass, like a list of things you will get. It became one sentence: responses vary widely, some people notice change quickly, others slowly, and some do not respond.
  • A saving. A course of treatment offered at a discount for paying ahead puts a price incentive in front of a clinical question. The savings language came off. The courses are still offered, with their terms discussed before anyone buys.
  • A comparison nobody measured. A table said other clinics typically charge more. Maybe so, but nobody had measured it. That cell now reads "Varies by clinic."
  • A date that would expire. A line about federal telehealth rules being in effect "through at least the end of 2026" was accurate on November 5 and would be stale by January 1. It now says care follows the rules in force at the time, and that some medications may need an in-person visit.
  • A door open early. Nine pages could be read as offering an office visit before the practice's new office opens. Every one now says the same thing: until the opening date, visits are by secure video.

The testimonials were a judgment call. The patient quotes on the providers page sound like outcomes, and the audit flagged them. The practice confirmed it has each patient's written permission, so they stayed, next to a disclaimer that was already better than most: "They describe an experience of care, not a result; results cannot be guaranteed."

Say only what is confirmed

One rule made the evening go quickly. If the practice had confirmed a fact, the copy could say it plainly. If it had not, the copy stopped asserting it, in either direction.

The clearest case was a medicine whose page used a word that implied a more standard kind of product than anyone had confirmed in writing. The tempting fix was to swap in the more precise word. But that would have been a new claim, made on my say-so. So the word came out and nothing replaced it, and the question went to the practitioner. Once she confirms it in writing, the page can say exactly what is true.

Diagram: a finding becomes a fix in four steps: the phrase and why a reviewer would stop; if the fact is confirmed, write it plainly; if not, remove the claim rather than assert the opposite; the Russian twin changes in the same release. Before and after: Research links this action to neuroplasticity, then Research in animals and early human studies links this action to neuroplasticity. All 201 sitemap addresses checked the same evening.
One exact phrase at a time, both languages together.

Each fix replaced one exact phrase. If the phrase on the live page no longer matched, because someone had edited it by hand since, the fix left that page alone and said so. The Russian twin of every page changed in the same release, because the site promises the same thing in both languages or it promises nothing.

Two things stayed back on purpose. New privacy and terms pages went up as drafts that cannot be published until every bracketed blank is filled and the practitioner and her attorney have read them.

What I would tell another practice

  • Read your own site the way a reviewer will. Before the fee, not after.
  • Find your best page and copy its habits. Most sites already have one that names its evidence and its limits.
  • Remove what you cannot confirm. Do not replace it with a different guess.
  • Watch dates. Any sentence that holds "through" a date is a sentence you have to come back to.
  • Change every language at once. A careful English page beside an older translation is still an old claim, just harder for you to read.

For the rules a facilitation practice adds on top of this, see marketing a psilocybin practice, and for how a licensed site is built around them, what a licensed practice site needs. The last check that night was a crawl of all 201 addresses in the sitemap, English and Russian, and every one of them answered.

Common questions

What do reviewers look for in health website copy?

Claims stronger than the evidence: results stated as likely, early research written as settled, unmeasured comparisons, savings attached to treatment, and dated statements that will go stale. The FTC standard is that health claims rest on competent and reliable scientific evidence.

Can a medical practice use patient testimonials?

Only if they are genuine, the patient gave permission, and they do not imply typical results. A clear disclaimer helps. On this site the quotes stayed because each patient's written permission was confirmed.

How do you rewrite a claim you cannot confirm yet?

Remove the claim and do not replace it with a different one. Put the question to the practice, and once the fact is confirmed, write it plainly.

Should a practice check its site before applying for LegitScript?

Yes. The application fee is paid before review and is not refunded if the answer is no. An evening of reading the site the way a reviewer would is far cheaper.

Related

← All insights