How we rate suppliers

12 categories, each scored out of 10, combined with the weights below (version 3). A supplier gets a PepFinder Score only once categories worth at least 50% of the weight have enough data. Until then its profile says “Not enough data yet” rather than showing a guess.

Where each score comes from

Every category score on a profile is labelled with its source:

• Automated data: what we read from the supplier's website and listed prices.
• User reviews: customers' star ratings for that category, once at least 3 have rated it. Five stars equals 10/10.
• Combined: both, blended. Reviews carry more weight as they add up; at 5 reviews they count equally with the automated data. The profile shows the split.
• Editorial assessment: set by our editors, with a public note saying what they checked.

“Would order again” percentages and customer-reported delivery times appear after 5 answers. Pros and cons are drawn only from customer ratings, once 5 customers have rated the same part of the experience; we don't write them ourselves.

Rankings use the same scores: best rated, best value, best selection, fastest delivery, best customer service, best payment options, best website, most reviewed. Suppliers without enough data for a ranking are left out, not ranked last.

A store earns a higher PepFinder Score by becoming more transparent, accurate and responsive, never by paying. Points come only when the underlying requirement is met, and they are checked:

• Complete, accurate details. Each detail counts once, however often it is edited. Details only the store states count for less until our crawler sees them on its website or PepFinder checks them.
• Verified store details. Checks PepFinder staff carry out themselves (control of the website, working contact channels, published policies) raise the Reliability score. They never involve naming the company behind a store, and a store that stays anonymous loses nothing. A paid plan or verification application is not an input.
• Rating shown on the store’s own website. A store that puts the free PepFinder badge on its site, so its customers see its live score, adds a point to Reliability. We count the badge only once it has loaded from the store’s own domain several times in the last month, and the badge is free on every plan.
• A catalogue that matches the website. A store's own product feed counts only where our crawler later reads the same price on its website. Supplier-sent prices are labelled and never enter price statistics.
• Replies and resolved complaints. Replies to reviews published since claiming, and complaints the customer (not the store) confirms as resolved.
• Performance figures. Dispatch times and similar figures a store reports are shown and labelled “store reported”. They move the score only once verified.

Every change to a score is recorded with its reason. Staff can freeze a score during an investigation or deduct points with a public note, which appears in the breakdown.

Score tiers: Excellent 9+, Strong 8+, Good 7+, Needs improvement 6+, and Limited information below 6. A tier describes how a store scores on PepFinder; it says nothing about the safety or quality of any product.

A separate, secondary score for readers who want the detail: what a supplier says about independent testing, whether it publishes certificates of analysis, and how open it is about how to reach it and where it ships from. It appears in the “Testing & transparency” section of each profile.

Criteria not yet assessed are shown as unknown, not as failures.

Certificates of analysis earn full credit only when we hold the certificates and have reviewed them. A supplier that links a certificates-of-analysis page we have not yet reviewed earns half; certificates over a year old earn less, and over two years less again. Conflicting evidence, or documents waiting for review, count as unknown until resolved.

Verification records what PepFinder staff have checked themselves. Each check has a date and an expiry, and only current checks count. Verification is never a statement that any product is safe or suitable for any use.

• Unverified: PepFinder has not yet checked this store itself. Everything shown comes from public sources and the supplier's own website.
• Website control checked: We confirmed that the people we dealt with control this website and its email domain. We never ask a store to name its company.
• Store checked: We confirmed website control, that the contact channels reach the store, and that its policies are published.
• Testing documentation checked: We checked a sample of the supplier's testing documents: batch-specific, dated, naming a real laboratory.
• PepFinder verified: Website control, contact channels, policies and testing documentation all checked by PepFinder, and current.

The share of the facts that matter about a supplier that we both know and can show a source for, weighted by importance. A fact counts only when its value is known (an explicit “no” is known; a blank is not) and a public source backs it. The facts and their weights:

Our crawler reads suppliers' own websites. Nothing it reads is published until it passes three steps. Validation: each field has rules; a “delivery time” that is really refund wording, or contains an email address, is rejected. Confidence: each value is scored by how it was found and whether other sources agree. Decision: high-confidence values are published as the supplier's claim; low-confidence values, and any value that differs from one already on the profile, wait for an editor. Rejected and held values are kept, with the reason, so every change can be traced.

Missing information is shown as unknown. We never treat “we did not find it” as “the supplier does not have it”.

Each listing is normalised: its form (vial, capsules, oral strips, spray, liquid, kit, blend), how many units it contains and the total mass. Only single-peptide vials with a clear price, currency and total mass are compared per mg: a 10-vial pack is compared on its 100 mg, and capsules, strips, kits and blends are listed but never compared with vials.

A price far from the rest of its market (judged on a log scale against the median, using the median absolute deviation, and at least 3× away) is held for review rather than averaged in. A price that jumps sharply between checks is held until we see it again. Market figures are shown only once at least 5 validated listings from 3 suppliers are available, and currencies are never converted or pooled.