Our rating methodology
178 cards in database Oldest live-card review
Sweepbase rates 178 crypto debit and credit cards using a 12-factor formula. Every score on the site can be reproduced from the published data and the rules below. There is no editorial discretion, no manual reranking, and no affiliate influence on the math.
Why a published methodology matters
Most crypto card comparison sites do not disclose how they rank cards. They publish ordered lists, claim to be "independent," and rely on readers to trust the order. In practice, those orders are often driven by affiliate commissions: the highest-paying issuer ends up at the top, and low-paying or non-paying issuers disappear from the list entirely.
We publish the formula because anyone should be able to challenge a rating. If our score for a card seems wrong, you can walk through the same factors using the public data on our card page and tell us where we are mistaken. Once an error is confirmed we correct it ahead of the next scheduled review pass.
Where the data comes from
Every field in our card database is collected from primary sources only. We do not aggregate from other comparison sites, scrape user reviews, or accept submissions from issuers without verification.
The sources, in the order we reach for them: the issuer's own fee schedule, terms of service and product pages; regulatory disclosures (FCA, FinCEN, EU MiCA filings) where a card has them; and press releases only when they come from the issuer's verified channels. 3 of the 178 cards are held and used directly; the rest are researched from those documents.
When a screenshot appears on this site, it was taken during that research, and the caption says what it shows: a page on the issuer's own site, a page of a fee schedule, a screen inside the app, or a transaction on a card we hold. Marketing renders supplied by an issuer are never captioned as evidence of testing.
The current database holds 178 cards with around 25 normalised data fields per card stored in our schema. The first full audit (April 2026) corrected 57 data points and removed 4 cards that no longer exist as standalone products. Since then the rows have been re-read against issuer documents in passes, and each card page prints the date of its own last review.
How we verify cards first-hand
Most of the database is researched from primary issuer sources. 3 of the 178 cards are held by an editor and spent on, and those carry a Tested hands-on badge. When we test a card first-hand, this is what we do, end to end:
Everything else is labelled for what it is. Four levels, and every card in the database sits in exactly one of them:
| Label | What it means | Cards |
|---|---|---|
| Tested hands-on | An editor holds the card and pays with it. | 3 |
| Verified against issuer docs | We read the issuer's own fee schedule, terms, and product pages. The card was never in our hands. | 166 |
| Pre-launch watch | Announced, but no customer has been issued one, so there is nothing to test yet. | 3 |
| No badge at all | Nobody has read a primary source for this card, so it makes no verification claim. | 6 |
The second row covers most of the database, so it deserves the same detail as the first. Reading a card's documentation means four specific things, every time:
- Open the issuer's fee schedule and terms of service, not the landing page
- Search that document for spread, stabilization, cross-border, and conversion, because that is where a “0% FX” claim usually gets its money back
- Compare the FX number in the terms against the number advertised on the product page, and record the conflict when they disagree
- Check the country list against the issuer's own restrictions page, then stamp the date on the card page as Last reviewed
The last row needs explaining. 2 of those cards can still be obtained today: Emoney, whose terms PDF answers 403 to every request we have made. Nobody has read it, so the card gets no badge rather than a badge it has not earned.
The rating formula
Every card starts at a base score of 2.5 out of 5. From there, we apply nine bonus factors and three penalty factors. The final score is clamped to the [2.3, 4.9] range and rounded to one decimal place.
Bonus factors
| Factor | Points | When it applies |
|---|---|---|
| Meaningful cashback | +0.5 | The cashback field describes a reward on purchases. A field that opens with "None", describes a suspended or discontinued programme, or describes fee discounts or a reward that is not yet live does not count. The formula does not weigh the rate or the tier needed to reach it |
| Self-custody | +0.4 | Funds remain in the user's wallet until the moment of purchase: the custody field opens with self-custody or non-custodial. A field that opens with "Hybrid" does not count |
| Premium card tier | +0.2 | The network field names a Signature, Platinum, Infinite, Black or Obsidian tier, or says Credit |
| Virtual + Physical | +0.2 | The Card Type field opens with both formats. A physical card mentioned later ("coming soon", a waitlist) does not count |
| Yield on deposits | +0.3 | The issuer runs a real yield product, either on the card balance itself or in a linked earn/staking product on the same platform. Fields that open with "No" or "None" never count, even when the rest of the sentence mentions APY |
| Low FX fee | +0.3 | The cross-currency rate is 1% or lower. The first clause of the FX field that states a rate counts, and a clause saying there is no FX fee counts as 0%; clauses about deposits, top-ups, loads, withdrawals, ATM use or staking, and fees the issuer says it removed, are skipped. A range counts at its upper end, and "not published" earns nothing |
| Zero issuance + annual fees | +0.2 | The entry card carries no issuance fee and no annual fee. A paid physical, replacement or extra card does not cancel the bonus; a "Free" that is gated (the field says it requires something, needs a subscription or minimum balance, is time-limited, or that a fee may apply) does not earn it |
| Apple Pay / Google Pay | +0.1 | The Mobile Pay field names Apple Pay or Google Pay as supported. A bare "Yes", or a wallet that is planned, rolling out or only reachable through another card, does not count |
| Live IBAN / SEPA rail | +0.2 | The IBAN/SWIFT/SEPA field affirmatively describes a working bank-transfer rail (personal or pooled IBAN, or a live SEPA path). It is not gated by whether SWIFT specifically is disclosed |
Penalty factors
| Factor | Points | When it applies |
|---|---|---|
| High FX fee | -0.3 | The cross-currency rate, read the same way as for the low-FX bonus, is 2% or higher |
| 3+ missing key fields | -0.4 | Three or more of cashback, issuance fee, annual fee, network, or regions are unavailable |
| 1-2 missing key fields | -0.2 | One or two of those key data fields are missing |
Deterministic noise
After bonuses and penalties are summed, we add a small deterministic noise value of roughly +/-0.1 derived from a djb2 hash of the card slug. This prevents dozens of cards from landing on identical rounded scores. Because the noise is deterministic, the same card always receives the same final rating. There is no randomness on each request.
Clamp and rounding
The final score is clamped to the [2.3, 4.9] range and rounded to one decimal. We do not award a perfect 5.0 because no card on the market is flawless across every dimension we measure, and we do not award scores below 2.3 because that would publicly shame products beyond what their data warrants. We simply de-prioritise or remove them.
How we rated the 1inch Debit Card, step by step
This is the full step-by-step calculation for the 1inch Debit Card, taken directly from our database. You can verify every value on the card's detail page. It shows the formula applied end to end. The card closed to new sign-ups in October 2026; the example stays because the arithmetic does not depend on whether you can still get it.
| Step | Delta | Running | Reason |
|---|---|---|---|
| Base score | 0 | 2.5 | Every card starts at 2.5/5 before applying any factors |
| Meaningful cashback (2% BXX / 1% BTC/USDT) | +0.5 | 3.0 | The cashback field names a rate on all card spend. The formula does not net it against the 2% card spend fee, which leaves the BTC or USDT option at about minus 1% |
| Self-custody check | 0 | 3.0 | Custody field reads "Custodial", so no self-custody bonus |
| Premium tier check | 0 | 3.0 | Mastercard Standard is not a premium tier, no bonus |
| Virtual + Physical formats | +0.2 | 3.2 | Both card formats are offered |
| Yield check | 0 | 3.2 | The yield field opens with "No": the live card page asserts no yield, and 1inch's lending products are credit against collateral rather than interest on the balance. No bonus |
| Zero issuance + annual fees | +0.2 | 3.4 | Both fees marked Free in our database |
| Mobile pay check | 0 | 3.4 | Mobile Pay reads a bare "Yes" that names no wallet, so no bonus |
| Live IBAN / SEPA rail | +0.2 | 3.6 | IBAN/SWIFT/SEPA field describes EUR and GBP bank transfers in and out of the account, a live rail, although no personal IBAN is named |
| FX fee check | 0 | 3.6 | The FX field opens with "No FX margin is published". The 2% card spend fee is a separate charge on the crypto spent, not an FX rate, so neither the low-FX bonus nor the high-FX penalty applies |
| Deterministic noise (slug hash) | +0.0965 | 3.6965 | djb2 hash of this slug yields +0.0965 (range is +/-0.1); reproducible, prevents identical scores from clustering |
| Clamp to [2.3, 4.9], round to 0.1 | 0 | 3.7 | Score is already inside editorial bounds; rounding to one decimal gives the published rating |
Final published rating: 3.7 / 5 (with deterministic noise, no clamp applied). The card is closed to new sign-ups; its page and full data stay at /cards/1inch-debit-card.
Editorial independence
Sweepbase earns revenue through affiliate partnerships with some card issuers. Affiliate status has zero effect on the calculated rating. The formula only reads objective attributes from our dataset, worth up to +2.4 / -0.7 points around the base score. Cards with a verified partner link are labelled “Affiliate partner” on their detail page. That label is a disclosure, not a recommendation and not a scoring input.
We list every significant crypto card we are aware of, including those with no affiliate relationship. If you find a card on the market that we have not covered, please email contact@sweepbase.net.
Read our full Affiliate Disclosure for details on which cards on Sweepbase carry affiliate links and how the commercial relationship is structured. Our Editorial Policy covers the full sourcing, verification, update, and corrections process in one place.
Audit and corrections
The first full audit of the Sweepbase database was completed in April 2026. Headline numbers from that audit:
57 data points were corrected across 145 cards (fee schedules, FX rates, regional availability, cashback claims) and 4 cards were removed as confirmed defunct or thin-content products. The rest stayed in the published database, which has grown since and holds 178 cards today.
If you spot a value that does not match the issuer's current published terms, please report it. Reader-submitted corrections are investigated as they arrive, ahead of the next scheduled pass.
Update cycle
Reviews land in passes: the catalogue is re-read a batch at a time, so review dates differ between cards, and each card page shows the month of its own last review. A confirmed fee or availability change is applied ahead of the scheduled pass, usually within a few days. The formula changes when an audit shows it misreads the data, and reader-reported corrections are investigated as they arrive.
Frequently asked questions
Formula last changed · Reproducible from public card data