A checklist for evaluating health claims online
You do not need a science degree to tell a substantiated claim from a sales pitch. You need eight questions and the willingness to ask them before you reach for a card.
1. What exactly is being claimed?
Write the claim down in one sentence. This sounds trivial and it is the step that does most of the work, because a large share of health marketing dissolves when you try to state it plainly.
"Supports healthy metabolism" survives as a phrase but not as a sentence: supports it how, measured how, compared to what? Vagueness is not sloppiness. It is a design choice that lets a reader hear a promise that was never made.
2. Is it a mechanism or an outcome?
"Ingredient X is involved in energy production in cells" is a mechanism. "Ingredient X will make you feel more energetic" is an outcome. The first can be entirely true while the second remains unproven.
Most ingredient-led marketing works by describing a real mechanism and letting the reader complete the sentence themselves. Notice when you are doing the completing.
3. Compared to what, and by how much?
A result without a comparison is not a result. Improved compared to no treatment, or compared to an existing option? A change of one percent or thirty? Over one week or two years?
Relative numbers deserve particular care. A risk falling "by half" is dramatic when the starting risk is 40% and nearly irrelevant when it is 0.002%. If only the relative figure is given, the absolute one is usually the less impressive of the two.
4. Who was studied?
Evidence in cells is not evidence in animals, and evidence in animals is not evidence in people. Studies in eighteen young male athletes may not describe a sedentary sixty-year-old. Findings in people who were deficient in a nutrient often do not transfer to people who were not.
If the studied population is never mentioned, assume it does not resemble you.
5. Who paid, and who benefits?
Funding does not make a study wrong, but it is information. So is the identity of whoever is telling you about it. A page that recommends exactly one product, sells that product, and mentions no alternative is doing something other than informing you.
Look for a disclosure. A site that explains its commercial relationships has told you how to discount it. A site with no disclosure and an obvious commercial motive has told you something less flattering.
6. Is the evidence linked, and does the link say what the page says?
Follow the citation. This is the step almost nobody takes and the one that reveals the most.
Common failures: the link goes to a homepage rather than a study; the study is in rats; the study used a dose ten times what the product contains; the study measured something else entirely; the study concluded that more research is needed, which the page rendered as "clinically proven".
7. What would it look like if this were false?
Legitimate claims can be wrong, and they tell you how you would find out: a measurement that could fail, a timeframe after which nothing happened. Unfalsifiable claims cannot fail. "It is working at the cellular level even if you feel nothing yet" is built so that no experience you have can count against it.
If nothing could ever disprove the claim, it is not a claim about the world.
8. What is the pressure doing?
Urgency is a sales technique, not a health fact. Countdown timers, disappearing stock, one-time prices and "they don't want you to know this" all exist to shorten the interval between feeling and paying, because reflection reduces conversion.
Real health information is not time-limited. Nothing true about your body expires at midnight.
Phrases that should slow you down
- "Miracle", "breakthrough", "secret", "banned" — theatre, not evidence.
- "Clinically proven" — proven by whom, on whom, showing what? Ask for the study.
- "Doctors hate this" — a claim about a conspiracy, not about a product.
- "100% natural, so no side effects" — a non-sequitur. Plenty of natural substances are potent, and potency cuts both ways.
- "Results guaranteed" — biological responses vary between people. A guarantee is a refund policy dressed as biology.
- "Lose X kg in Y days" — a specific amount in a specific timeframe is the shape of claim regulators treat most sceptically, and for good reason.
- "No diet or exercise required" — claims of effort-free change are among the most reliable markers of an unreliable seller.
What before-and-after photographs actually show
They show two photographs. Lighting, posture, timing, clothing, hydration and camera angle can produce a striking difference within a single afternoon. Even when a transformation is entirely genuine, it is one person's experience, selected precisely because it was unusual — the pictures that were not remarkable were not published.
Individual testimony is not evidence of what a product does in general. That is not scepticism about the person; it is arithmetic about selection.
The two-minute version
- State the claim in one plain sentence.
- Decide whether it is a mechanism or an outcome.
- Look for the comparison and the size of the effect.
- Check who was studied.
- Find the funding and the disclosure.
- Follow one citation and read the conclusion.
- Ask what would falsify it.
- Notice the urgency, and subtract it.
Then take what is left to a pharmacist or doctor. They will tell you in thirty seconds whether it interacts with anything you already take — which is the question the sales page will never raise.