The four blocks
Every score out of ten is built from four weighted blocks. The weights shift slightly by category — a hardware wallet is not scored on order-book depth — but the blocks are the same, so a score in one category means roughly what it means in another.
- Cost (30%): published maker and taker tiers, the spread markup on a simple buy, deposit and withdrawal charges, network fee handling, and what the advertised cheap rate actually requires in volume or token holdings.
- Custody and solvency (30%): who controls the keys, which legal entity faces the customer in each region, what reserve evidence exists, how recent it is, and whether the audit scope covers liabilities as well as assets.
- Depth and execution (20%): liquidity in the pairs people actually trade, routing and slippage behaviour, uptime during volatility, and the quality of the order types on offer.
- Getting out (20%): withdrawal limits and processing times, fiat rails, documented account-closure behaviour, and the verification friction that appears at withdrawal rather than at signup.
What evidence counts
Primary sources first: the platform's own fee page, terms of service, security documentation, status page and regulatory filings. Where a figure comes from a third party — an attestation firm, an analytics provider, a regulator's register — the review names the source.
- Marketing claims are reported as claims, with attribution, never as verified fact.
- A screenshot in a review shows the page as it appeared on the capture date.
- Where a platform publishes different terms per region, the review says which region it is describing.
- Figures that could not be verified are described as unverified rather than omitted.
What lowers a score
Opacity is penalised as heavily as bad pricing. A platform with mediocre fees that publishes them clearly will outrank one with attractive headline fees buried behind a login wall.
- Fee schedules that are not reachable without an account.
- Reserve attestations older than a year, or ones that cover assets but not liabilities.
- Withdrawal terms that differ materially from the signup experience.
- Regional entity structures that make it unclear who holds customer assets.
What we will not do
We do not publish price predictions, buy or sell calls, yield recommendations, or anything that could be read as personalised financial advice. We do not score a product we could not reach a working public website for. We do not accept payment to review, re-review, re-rank or remove a product.
Updates and re-reviews
Crypto platforms change fees, custody arrangements and regional availability frequently. Every review carries its publication date, and material changes trigger a re-review rather than a silent edit. When a score changes, the reason is stated in the review.
Independence
Editorial decisions are made on editorial grounds. Reviewed companies get no advance approval over coverage, no right to see a draft, and no ability to remove criticism. Where a writer holds a position in an asset or has any relationship with a reviewed platform, they do not write or edit that review.