Identify the product.
A new SKU enters the queue when (a) we receive a tip, (b) a reviewed brand launches a new product, or (c) it appears in our quarterly shelf-survey of major Indian retailers and online platforms. Each entry gets a unique product ID and is linked to a brand record.
Source the unit.
We buy the actual unit anonymously from a normal retail or D2C channel — never from the brand's PR contact. We document the channel, batch number, manufacturing date, MRP, and shipping date so that purity follow-ups have a referenceable batch.
Enter label data, blind.
The supplement-facts panel, ingredient list, and any claim-bearing copy on jar/box/insert are entered into the database — without the brand name visible to the data-entry researcher. We record dose per serving, form, extract identity (e.g. KSM-66 vs generic root), and supplier where stated.
Verify against supplier docs.
We request batch-level Certificates of Analysis from the brand or from the upstream supplier. If documentation is unavailable, the product is flagged "unverified supplier" — that flag affects the Purity sub-score directly.
Commission lab work where it matters.
For categories where label vs reality routinely diverges (fish oil freshness, whey protein content, heavy-metal contamination in herbal extracts) we commission independent lab analysis at NABL-accredited facilities. Lab partners do not see brand identity.
Score against the rubric.
The reviewer scores the entry against five dimensions — clinical dose, ingredient form, purity, value, and label honesty — each on a 0–10 scale with a written standard. The rubric weights are fixed and public; see the rubric page.
Second-pair-of-eyes review.
Every score is reviewed by a second researcher before publication. Disagreements over half a point trigger a panel review with three reviewers. The published number is the panel median, not the original draft.
Publish, with the citations attached.
The review goes live with every cited paper, every COA, and every lab report linked from the Sources section. Subscores are visible — readers can see which dimension dragged the total up or down.
Re-review on triggers.
A published score is automatically re-opened when (a) the brand changes formulation, (b) a new batch tests significantly outside the prior range, (c) we receive credible reader reports, or (d) annual review at the 12-month mark. Every re-review is logged in the database changelog.
The shortcuts we refuse to take.
No press samples.
We don't accept anything for review. Every unit is purchased; the cost is part of our editorial budget.
No "best of" rankings for pay.
Lists on Naked Compound are the result of database queries against the rubric. Nothing on this site can be paid into a list.
No anecdotal scores.
If a sub-score can't be tied to a measurement, document, or written rubric standard, it doesn't get included.