How The Milligram tests and scores a supplement
We buy every product we review, take it for at least eight weeks, score it against eight fixed criteria and publish the criterion that drags the average down along with the ones that lift it. This page is the full method, including the parts that are limitations rather than strengths.
Who we are
The Milligram is a small independent review desk published by Milligram Editorial Group, LLC in Corvallis, Oregon. We started in 2019 for a boring reason: one of us went looking for a straight answer about a supplement a relative had ordered and found nothing but sales pages wearing a review costume. The desk covers the direct to consumer end of the supplement market, which is the corner with the loudest claims and the least shelf level scrutiny.
Reviews are written by the research desk and signed off by Tessa Ordway, MS, RD, LD, a registered dietitian who handles the ingredient and safety sections and has the final say on the score. We are not physicians, we do not run a laboratory, and nothing we publish is medical advice or a substitute for your own clinician.
How we buy
We buy at retail, through the same checkout a reader would use, with our own money. We do not accept free samples, we do not take payment for a review, and no brand sees a draft before it is published. If a manufacturer contacts us after publication with a correction we check it and fix what is wrong, in public, with a note.
The test protocol
Every product gets at least eight weeks and at least two testers. For Sugar Reset we ran ten weeks with two people on the label dose of one dropper each morning. We log the first week separately, because that is where format problems appear, then log weekly through to the end.
What we can honestly claim from this: a real account of what taking the product is like, what the format gets right and wrong, whether the claimed effects show up in ordinary use, and whether anything unpleasant happens. What we cannot claim: statistical significance, blinding, a control group or laboratory verification of the contents. Two people over ten weeks is a structured experience report. Anyone presenting that as clinical evidence is selling you something.
The eight criteria
Every review is scored on the same eight criteria, each from 1 to 5 in tenths. The published score is the unweighted mean, rounded to one decimal. We do not weight the criteria, because weighting is where a reviewer hides a thumb on the scale.
| Criterion | What it measures |
|---|---|
| Effectiveness | Whether the claimed effect shows up for our testers and in the reader pool, and how large it is |
| Value for money | Cost per day at the cheapest honest pack size, against comparable products in the same niche |
| Ease of use | How likely a normal person is to keep taking it for eight weeks |
| Onset speed | How long before anything is noticeable, measured against what the marketing implies |
| Tolerability | Side effects, stomach comfort, and how often problems appear in reader reports |
| Taste | For anything you have to taste. Capsule products score this on swallowability instead |
| Label transparency | Whether the label prints amounts, whether the site matches the bottle, and whether a clinician could assess it |
| Repurchase intent | Whether people buy it again, from the reader pool rather than from our own panel |
A product that hides its doses can still score well overall, because five of the eight criteria are about the buying and taking experience. It cannot score well on transparency, and we will not average that failure away quietly.
How we collect reader ratings
The reader figures on our reviews come from public reports we collect by hand: ratings and written comments that readers and buyers have posted publicly about the product. We read each one, tag the themes it mentions, record the star rating, and note whether it references a purchase. We do not buy review data, we do not scrape at scale, and we do not include anything a brand supplied to us.
The limitations are worth stating plainly. Public supplement reviews skew towards people with a strong opinion in either direction. We cannot verify a stranger's purchase independently, so our verified share reflects reports that reference an order rather than a confirmed transaction. Percentages are rounded, so a column may total 99 or 101 before rounding is applied. These numbers are a fair summary of what the public record says, not an audited dataset.
Reviews submitted through the form on our review pages are read by the desk, but they are not added automatically to any published figure and they never appear in structured data. A number that cannot be checked does not belong in a schema block.
How we stay independent
The Milligram is funded by affiliate commissions. When a link on one of our pages sends a reader to a manufacturer's official website and that reader buys something, the manufacturer may pay us a percentage. The reader pays the same price either way.
Three rules keep that from becoming a problem. Scores are agreed and written down before any affiliate link is added to a draft. Commission rates are never known to the person writing the score. And we publish the criteria that make a product look bad in the same table as the ones that make it look good, which is the only structural defence a funded review site actually has. Read the full terms on our affiliate disclosure page.
Questions, corrections or a complaint about something we published: write to [email protected] and put the product name in the subject line.