How we rate

• Independent. We do not sell any product we rate.
• Source-backed. Every score traces to data you can check — we show the per-source breakdown on each page.
• Unbiased by money. Sponsorship affects placement labels only, never the score.

Each source carries a trust weight — regulatory data outweighs crowd sentiment, which outweighs unverified opinion. To stay honest, the table below lists only the sources currently feeding live scores (with how many data points each contributes). Other sources are weighted as shown the moment they come online.

Today scores are driven by FDA openFDA regulatory data (recall history). As marketplace and moderated user reviews accumulate, they join the weighted average automatically.

We compute a single composite from the per-source ratings using a trust- and confidence-weighted average, then present it on a 0–100 scale:

1. Shrinkage.Each source's raw score is pulled toward an industry baseline of 3.5/5 (70/100) based on how many data points back it: a source reaches roughly half its full weight at ~50 data points, so thin evidence stays near the baseline.
2. Weighting. When more than one source is available, each is weighted by its trust level multiplied by its data volume, and we take the weighted mean. A product with no data shows as Not yet rated rather than a made-up score.
3. Transparency.The contributing sources — and their data-point counts — are shown on every product and company page under “How we rate”.

Scores run 0–100. For quick scanning we color the score chip: 70+ strong, 50–69 mixed, <50 concerning.

Visitor reviews are moderated before they appear or affect any score. Unverified submissions never move a rating. This protects against spam and manipulation — important for health-related content.

Do you sell any of the products you rate?

No. pharmaranks is a ratings and directory service only. We never sell drugs, supplements, or devices. Some outbound links to partners are affiliate links, but they never influence a rating.

Can a company pay for a better rating?

No. Sponsored placement is labeled and only affects listing position, never the composite score. Scores are computed automatically from data sources and cannot be bought.

Why does a product with few reviews not have a high score?

We apply Bayesian shrinkage: a rating backed by very few data points is pulled toward a neutral baseline until enough evidence accumulates. One five-star review does not outrank a long track record.

Is this medical advice?

No. Our ratings and information are for general reference only and are not a substitute for professional medical advice. Always consult a licensed healthcare provider.

Where does the safety signal come from?

From the FDA's openFDA enforcement (recall) data. Each product starts from its manufacturer's recall record, then drops for its own FDA recall history — a product that was itself recalled scores below a clean product from the same maker. Products with no recall history of their own reflect their manufacturer's record. We use recalls rather than raw adverse-event report counts, which are skewed by sales volume. Recalls are weighted by the FDA's own severity class rather than counted equally: Class I (a reasonable probability of serious harm or death) counts fully, Class II (temporary or reversible harm) half, Class III (not likely to cause harm) a fifth. We do not weight by our own classification of what a recall was about — that reads the recall's reason text, which we hold for roughly a tenth of the recalls in the FDA database, so scoring by it would rank products by what we happen to store rather than by what happened.