How we rate crypto cards
One formula for every card. Version 1.0, 16 October 2026.
Each card gets a score from 0 to 10. The formula reads only the card's data: fees, rewards, limits, availability and so on. It does not read affiliate links, commissions or the editor's pick, and a test in the code checks that changing them does not move any score.
What counts, and how much
Ten criteria, each scored 0 to 10, then weighted. The weights are ours: no provider or partner set them. This is why each one has its share:
| Criterion | Weight | Why this weight |
|---|---|---|
| Fees | 28% | What you pay to get the card, keep it, load it and spend with it. Fees differ more between cards than anything else, and at ordinary spend they decide more money than cashback does, so they carry the most weight. |
| Rewards | 18% | Cashback is real money back, but it is capped, banded and often paid in a volatile token. We score it at $1,000 of spend a month on the free or cheapest plan, the level most people reach. |
| Availability | 14% | A card you cannot get is worth nothing to you, and most people pay by phone, so countries served and Apple Pay or Google Pay support count. |
| Limits | 7% | High limits matter to heavy spenders and travellers who need cash; most people stay well below them. |
| Custody | 7% | Who holds your money decides who can freeze it. Self-custody scores highest, custodial lowest. |
| Track record | 7% | Time on the market, a named card issuer and a clean incident record. |
| Data completeness | 6% | Terms a provider does not publish are a cost too: you find them out after you pay. |
| KYC and privacy | 5% | Skipping ID checks matters to some people. The weight is small on purpose, so that skipping KYC alone cannot lift a card; its price shows up in the other criteria. |
| Networks and coins | 5% | More networks and coins make funding easier and cheaper. |
| App ratings | 3% | Store ratings are noisy and sometimes rate a whole wallet app, so they count least. |
| Total | 100% |
The formula
Most sub-factors are compared across all 26 cards as a percentile: a card scores 10 times the share of cards with a different known value that it beats. Equal values are left out of the comparison, so a value shared by the best cards scores 10 and one shared by the worst scores 0. Lower is better for fees; higher is better for cashback, countries, limits, networks, coins, age and store ratings.
A few sub-factors use fixed points instead: custody (self-custody 10, partly self-custody 6, custodial 3), identity checks (no ID at signup 10, wallet without KYC 7, KYC required 4), wallets (5 each for Apple Pay and Google Pay, 1 when unknown), reward currency (stablecoin or USD 10, bitcoin 8, other tokens 5), a named issuer (10, otherwise 2) and incidents (none 10, any 4). Data completeness is the share of 20 key fields the provider discloses, times 10.
Each criterion score is rounded to 0.1. The total is the sum of weight times rounded score, rounded to 0.1, so you can recompute it from the table on any card page.
Fixed fees become percentages on typical amounts: a $500 load, a $50 purchase, a $200 ATM withdrawal and $1,000 of spend a month.
criterion score = round(Σ sub-weight × sub-score, 0.1)
total = round(Σ weight × criterion score, 0.1)
percentile score = 10 × beaten / (beaten + beaten by)Sub-factors inside each criterion
- Fees: Cheapest card 20%, Yearly cost 20%, Top-up cost 20%, Cost per purchase 15%, FX cost 15%, ATM cost 10%
- Rewards: Cashback at $1,000 a month 75%, Reward currency 15%, Perks 10%
- Availability: Countries 70%, Apple Pay and Google Pay 30%
- Limits: Spending capacity a day 70%, Cash a day 30%
- Networks and coins: Networks 50%, Coins 50%
- Track record: Age 50%, Issuer 30%, Incidents 20%
Unknown values
When a provider does not publish a value we do not treat it as zero, and we do not use the average. An unknown sub-factor scores 2 out of 10, as if the card sat at the 20th percentile, and it also lowers the data completeness criterion. Store ratings with fewer than 20 reviews are shown on the card page but not scored. On the site unknown values say Not disclosed or Unknown; they are never shown as 0 or N/A.
No-KYC cards
Cards that issue without an ID document at signup go through the same formula; we did not change it for them. They get the full 10 for identity checks, but that criterion weighs only 5%. What skipping KYC costs shows up elsewhere: they are custodial, loading costs 2.8% to 5.49%, the projects are young, issuers are not named, and limits and country lists are often unpublished, which lowers availability, limits, track record and completeness. That is why they rank in the lower half of the catalog. Their pages and the no-KYC hub explain the trade-offs.
Sources and their priority
We take figures from the provider's own website, help centre, fee pages and terms first. Where those are silent we use DefiLlama, then independent reviews and guides, then Telegram reports, which are the least reliable. When sources disagree and the conflict cannot be settled, we use the cautious value (the higher fee, the lower reward) and list both on the card page. Figures not confirmed by the provider carry an asterisk. Aggregators are never preferred over the provider.
Which cards we include
We cover consumer crypto cards that are live and open to new users in at least some countries. Cards that have closed or are closing, business-only products and cards with very little use are left out. For no-KYC cards we also require published fees and terms and no widespread reports of frozen funds, missing cards or fraud. Exchange cards are a separate group we plan to add later.
Editor's choice
One card carries the Editor's choice mark: ether.fi Cash. It is an editorial pick based on our own use of the card, made separately from the rating. The mark appears next to the card in the catalog and on its own page; it does not change the score or the order of the catalog. Some cards, including the one picked, may have an affiliate link; see how we make money.
Money and independence
The Order card buttons are affiliate links and we may earn a commission. The rating does not read anything about commissions: the same data produce the same score whether a card has an affiliate link or not. Affiliate disclosure.
Version
Rating formula 1.0, last changed 16 October 2026. When the formula changes, the version number changes and this page explains what moved.