Ratings methodology

We collect public review data from independent platforms. We do not host reviews ourselves, and we do not allow firms to submit their own scores.

New sources are added over time. When we add or adjust a source's weight, every firm's rating is automatically recalculated.

Not every source is equally reliable. A verified-purchase review on Trustpilot carries different signal than an anonymous Discord post. Each source has a base weight from 0 to 10 that reflects its trustworthiness, anti-fraud measures, and signal-to-noise ratio.

Within a source, an individual firm's contribution to the aggregate is also scaled by its review count on that source — a firm with 4,000 Trustpilot reviews contributes more to the weighted average than a firm with 12.

We use a Bayesian-weighted average to prevent firms with very few reviews from gaming the top of the rankings. The formula:

weighted_rating = (C × m + Σ(rating_i × review_count_i × source_weight_i))
                  ─────────────────────────────────────────────────────────
                       (C + Σ(review_count_i × source_weight_i))

where:
  C = 50      (prior strength — how many "neutral votes" to assume)
  m = 3.5     (industry mean prior — neutral baseline rating)
  i           iterates over every (firm × source) data point

In plain English: a brand-new firm with no reviews starts at 3.5 (neutral). As real reviews arrive, the rating gradually moves toward the actual community score. A firm needs substantial review volume — weighted by source trust — before its rating diverges meaningfully from neutral.

Every firm's rating is paired with a confidence tier based on its effective review weight — the sum of (review_count × source_weight) across all sources.

Source data is refreshed continuously by our collectors. When a new data point arrives — for any firm, from any source — the affected firm's aggregate rating recalculates automatically within seconds, no manual intervention required.

Each firm profile shows the last calculated timestamp. Individual source rows show last scraped times so you can see how fresh each input is.

• No firm can pay to alter its rating, increase its weight, or remove individual reviews.
• We don't write reviews ourselves and we don't allow firm-submitted reviews into the aggregation.
• Source weights are set on industry-wide criteria (anti-fraud, verification, sample size, transparency) — never per firm.
• If we remove or add a source, every firm in the directory is recalculated under the same rules.
• The Bayesian prior (C = 50, m = 3.5) applies uniformly to every firm.

Every component of every rating is publicly inspectable on each firm's profile: which sources contributed, their raw scores, their review counts, their weight, the percentage they contributed to the final aggregate, and a link out to verify on the source platform.

If you find a data point that looks wrong, the linked source URL is the source of truth — if it has changed, our next scrape will reflect it.

Questions or corrections: contact us.