Methodology
This page sets out, step by step, how PlatformsReviews collects reviews, cleans them, computes a rating and orders its best-of guides. It exists so that any score or ranking on the site can be traced back to a stated rule rather than an opinion.
1. Sourcing reviews
Reviews are submitted by users through the profile forms. Historic reviews imported during the original data build are labelled as imported and held to the same guidelines as new submissions. No review is authored by PlatformsReviews staff, and no company writes its own listing text into the review set.
2. Cleaning the data
Before a review counts it passes automated and manual checks. Recovery-scam, refund and contact-harvesting content is stripped. Reviewer identities that were stored as raw email addresses are reduced to a display name or to a neutral label, so no email is exposed. Duplicate bodies repeated across multiple companies are treated as one suspicious pattern and excluded.
3. Computing the rating
The company rating is the arithmetic mean of the surviving verified star ratings. The review count is the size of that surviving set. The star graphic, the numeric average, the aggregate rating in structured data and the card figures on homepage and category pages all derive from this single computation, so no two views of the same company disagree.
4. Ranking the best-of guides
The best-of guides rank companies within a platform type by combining two measured signals: the average rating and the number of reviews behind it. A high rating carried by a large sample is placed above an equal rating from a thin one. Where a guide names a strength for a company, that strength summarises recurring themes in the review bodies for that company. It is not a claim by PlatformsReviews and never a fabricated fact.
5. What is never invented
PlatformsReviews does not publish licence numbers, regulator identifiers, fee figures or founding dates unless they appear on a profile as user-submitted or company-stated, in which case they carry a not-verified label. No named analyst persona rates companies. Where authorship is shown, it is the organisation itself.
6. Corrections and removals
A review is removed if it breaches the guidelines, not because it is critical. Companies may dispute a review or request a factual correction through the contact page. Every removal triggers a recomputation of the affected score.
7. Freshness
Scores recompute whenever the underlying review set changes. Best-of guides therefore reflect the live corpus rather than a frozen snapshot. This page carries the date of its last substantive revision.
Last reviewed: 2026-09-28.
Related: how we rate, review guidelines, content moderation, editorial policy.
Frequently asked questions
What is the PlatformsReviews methodology in one line?
Collect verified user reviews, remove spam and duplicates, average the surviving star ratings, and rank best-of guides by rating and review volume together.
Are any facts invented?
No. Licence numbers, fees and dates appear only when user-submitted or company-stated, and are labelled as not verified. No fake analyst rates companies.
How are best-of rankings ordered?
By average rating combined with review count, so strong scores backed by many reviews outrank the same score from few.
When was this methodology last updated?
The page shows a last-reviewed date and is revised when the process changes.
Who writes the reviews?
Users. No review is authored by PlatformsReviews, and companies cannot write their own review text.