Almost every overstated claim in this industry started life as a real study. The study was usually fine. What happened to it on the way to a product page was not.
Here is a practical way to check a claim, using a real paper as the worked example.
1. Find the actual paper
Not the press release, not the article about it, not the LinkedIn post. If a claim cannot be traced to a paper with authors, a journal and a DOI, that is the finding.
Our worked example: Wang et al., Journal of Clinical Sleep Medicine 2025;21(1):69-80, a double-blind, placebo-controlled, randomized crossover study of a CBD and terpene formulation in insomnia. Real paper, real design.
2. Separate enrollment from analysis
These are different numbers and the larger one is the one that gets quoted.
That trial enrolled 125 people. It analysed 56. Roughly half dropped out, which the paper acknowledges reduced statistical power. Anyone describing it as “a study of 125 patients” has read a summary rather than the paper.
Always ask how many finished, not how many started.
3. Check how the outcome was measured
Measurement method sets the ceiling on what a study can claim.
In that trial, sleep stages were measured with a wrist-worn consumer device rather than polysomnography. Actigraphy infers sleep stages from movement and heart rate. It does not measure brain activity. For a decentralized study at that scale it is a defensible choice, and it still means the results describe what a wearable detected.
Ask what instrument produced the number, and what that instrument can actually see.
4. Compare the objective and subjective results
This is the most revealing check and it is almost never reported.
In that trial the device detected a mean 1.3 percent increase in time spent in slow-wave plus REM sleep, 95% CI 0.1 to 2.5, P = .03. The paper calls it marginal. Self-reported sleep measures showed no statistical significance.
So the instrument saw a small change and the participants did not notice one. That is not a scandal, it is a normal and honest result. It is also the difference between “may affect sleep architecture” and “helps you sleep better”, and only one of those is supported.
5. Read the funding statement
That study was funded by Defined Research, Inc., and multiple authors hold equity or stock options in the sponsor.
Industry funding does not make a study wrong. Most applied research in this field is industry funded, and the alternative is often no research at all. But the disclosure is part of the finding, and it is the line most often deleted in the retelling.
6. Check what the design can attribute
The subtlest error, and the one that produces most terpene marketing.
The formulation in that trial contained 300 mg of CBD plus eight terpenes at 1 mg each. There was no CBD-only arm. So the design cannot separate what the terpenes contributed from what the CBD contributed. Any claim that the study shows terpenes improve sleep is attributing an effect the design cannot isolate.
Ask what was compared with what. If a combination was tested against placebo, the finding belongs to the combination.
7. Notice the species and the model
A great deal of terpene research is cell-based or animal-based. That work is genuinely valuable for understanding mechanism and it does not transfer directly to humans, particularly at concentrations a person would ever encounter.
The tell is a paragraph that opens with a cell study and closes with a sentence about what it means for people. Watch for the moment the subject quietly changes.
Why bother
Commercially, because accuracy is the cheapest differentiator available in a category where nearly everyone overstates. A buyer who catches you being scrupulous about a study you could easily have exaggerated will believe you about the things they cannot check.
Practically, because overstated claims are also the ones that attract regulatory attention. The discipline that keeps you honest is the same discipline that keeps you out of trouble.
If you want the current state of play on the biggest open question in this area, the entourage effect evidence is a good place to practise all seven checks at once.
About this article
Written 31 July 2026 and checked against the statute, the Federal Register and the published papers rather than against secondary coverage. Regulatory positions in this area are moving, so check the date before relying on it. Nothing here is legal, tax or medical advice.
