Methodology: where every number comes from
Every figure on this site carries an evidence tier — issuer-documented, modelled, or editorial — and the score behind every card is the sum of published parts. This page shows all of it: the tiers, the formula, the pipeline, and the things we deliberately do not claim.
The three evidence tiers
A number is only as good as its source, so every entry on the site is tagged with one of three tiers and the tag travels with the figure wherever it is shown.
| Tier | What it covers | Volume in the dataset |
|---|---|---|
| Issuer-documented | Fees, KYC rules, networks and card types read from the issuer's own published pages, with the source URL kept on file. Checkout attempts recorded against issuer documentation. | 6 BIN ranges 1,474 attempts |
| Modelled | Checkout behaviour estimated for a BIN range from how cards of the same network and card type perform across the log. Published as an estimate — never as a test we ran. | 417 BIN ranges 90,435 attempts |
| Editorial | Judgement calls by a named editor, with the reasoning shown on the page rather than hidden behind a score. | Every verdict and ranking |
Brand profiles carry the tag as a Data quality row: a profile marked "partial — terms reviewed, no checkout log" is telling you exactly which tiers it rests on. 91,909 attempts and 65,899 cleared charges are on file in total; the split above is the honest version of that headline number.
The composite score, itemised
Each card's score out of 100 is the sum of the parts below — nothing is weighted behind a curtain. If a component cannot be read from documentation, it scores the published fallback and the profile says so.
| Component | Range | What it reads |
|---|---|---|
| Base | 36 | Starting point for every tracked card. |
| Data quality | 0–10 | How much of the profile is backed by issuer documentation rather than a crawl. |
| KYC depth | 0–9 | No KYC 9 · minimal or ID-only 7 · unpublished 3 · face verification 0. |
| Card creation fee | 0–9 | Free 9 · up to $3 8 · up to $10 6 · up to $30 3 · above that 0. |
| Deposit fee | 0–9 | Free 9 · up to 1% 8 · up to 3% 6 · up to 5% 3 · above that 0. |
| FX fee | 0–4 | Up to 1% 4 · up to 2% 3 · up to 3% 1 · above that 0. |
| 3-D Secure | 0–6 | Supported 6 · not supported 1 · unpublished 2. |
| Risk level | 0–8 | Low 8 · low-medium 6 · medium 4 · medium-high 1 · high 0. |
| Checkout coverage | hits × 1.3 | Number of the 105 use-case pages the card is documented to clear. |
| Billing address | 0–3 | Custom address 3 · US general address 2 · none 0. |
| Alipay & WeChat | +4 | Funding through either channel is supported. |
| Cards per account | +2 | Unlimited cards on one account. |
| Category | −5…+2 | Anonymous −5 · B2B / prepaid / traditional −2 · fintech +1 · crypto +1 · neobank or digital bank +2. |
| Crawl-only data | −2 | Profile built from a crawl or search rather than documentation. |
| Cap | 0–100 | The composite score is clamped to this range. |
From composite score to the 10-point rating
Scores are relative, not absolute. A card's rating out of 10 is its percentile position across the tracked set, mapped as:
where p is the card's percentile in the composite ranking. The curve is deliberately steep at the top, so a 9.0 means top decile — not "nine out of ten in the abstract". Stars show the same rating in halves.
The six-axis module on brand pages
Every brand profile carries a six-axis breakdown so you can see why a score landed where it did, not just the total:
- 1Card fees — creation, monthly, top-up and FX, read from published pricing.
- 2Checkout coverage — which of the 105 use cases the card is documented to clear.
- 3Verification — KYC depth and what identity documents are asked for.
- 4Risk & safeguards — issuer risk level, 3-D Secure, freeze controls.
- 5Funding & setup — top-up channels, billing address handling, cards per account.
- 6Tracked clear rate — the checkout evidence behind the range, shown separately because it feeds the log, not the composite.
How the dataset is assembled
One structured dataset feeds every page, which is why the numbers agree across the site: an issuer's fee shows the same on its profile, in the country guides, in the comparison pages and in the card finder. The pipeline:
- 1Read — fees, KYC, networks and card types from each issuer's own published pages.
- 2Structure — rows normalised into one comparable schema, currency converted at the review-date rate.
- 3Score — the composite formula above, plus the six-axis breakdown.
- 4Publish — 484 issuer profiles, 105 use-case pages, 423 BIN ranges, 23 country guides, and a 484-entry card finder — all from the same source.
What we don't claim
- ×No permanent BIN labels. A prefix is shared by every card an issuer produces — a strong hint about who is behind a card, never a guarantee.
- ×No modelled number is a test. Where a figure is modelled, the page says so; the raw attempt log sits in the BIN lookup.
- ×No permanence. Acceptance shifts by the month, sometimes by the day, and it varies by billing region. Treat figures as a starting point for your own checks.
- ×No financial advice. Nothing here is a recommendation to open a card with any specific issuer.
Update cadence
Fees are re-checked against issuer pages on a rolling monthly schedule. Every page carries a "last reviewed" date so you can judge freshness at a glance. Country guides publish a per-pass changelog — what moved, and when. If you spot a price that has changed or a source worth adding, write to [email protected] and it goes on the re-check list.