Only values with a source
We only take values a provider publishes itself. No source, no value.
- For every value we store the source, the verbatim quote and a copy of the page. The number has to be in the quote.
- The small s next to a value shows its source and check date.
- If a value is missing, it says "not published", never 0 or an estimate.
What providers send us
Details that providers only send to us are not used. What is not publicly available cannot be checked by anyone.
- A provider who wants a value corrected sends us the link to the place on its official website.
- There are no special rules for providers, with or without a contract. We list everyone, including doubtful providers.
How we check
Our goal is the highest possible data quality and the fastest reaction when a provider changes something. With thousands of source pages, no editorial team can do that by hand. So AI agents read the providers' pages, and fixed rules in the system make sure nothing wrong gets published.
- source pages monitored
- 3,567
- of them checked in 24 hours
- 2,694
- values with a source
- 12,597
- changes in 30 days
- 1,463
- A second AI model from a different maker reads important values again, blind. They only appear if both agree.
- Unusual values are held back until the second check confirms them.
- New providers are added by a human.
Nothing disappears quietly
Every value is versioned. Old values are not overwritten but kept with their date and source.
- Every product page shows its changes with the old and new value, including ended promotions and incidents.
- Every change is written to a log that cannot be altered afterwards without that showing.
How we examine providers
A good fee table says nothing about the company behind it. So we check that too, the same way for every provider – the hardest part of our work.
- Domain age from the official registry.
- Web archive analysis: since when today's operator has really been online. This reveals bought domains with borrowed age.
- Licences, hacks, halted withdrawals and regulator warnings, each with a source.
- Ratings from app stores and Trustpilot count only in damped form.
How we make money
Collecting and checking data costs money. We earn it in two ways, and both are labelled on the page.
- Partner links: we have a partner agreement with some providers, by no means all. Links to those providers are affiliate links and labelled as such. If you sign up through one, we may receive a commission.
- Sponsored slots: providers can book a sponsored slot. Their card then also appears in a labelled area above the list.
- Bonuses and promotions we show are the provider's own offers with a source, not advertising by us, and carry no partner link of ours.
What a sponsor gets
- a slot above the list, labelled "Sponsored"
- nothing else: its card stays at its normal position in the list
Not for sale at any price
- different or embellished values
- a better score or a higher position in the list and filters
- removal of warnings, incidents, poor values or old versions
- exemptions from sourcing and the second check
- influence on texts, news or the early warning
- a preview or sign-off before publication
The same applies to partner links: whether there is a contract changes neither values nor score nor order.
Score and order
The score combines sourced values according to a fixed, published methodology. Criteria and weights are on the methodology page of each category.
- The score is calculated from the data, not assigned by hand, and cannot be bought.
- The default order follows the score. Sponsored slots and partner links move nothing.
- When we change the methodology, the version number published with the weights goes up.
- The score is a methodical comparison, not a recommendation for your individual case.
What applies in which category
Sourcing, the second check, the history and the domain and web archive analysis apply everywhere. Which checks and notifications are added depends on what makes sense in the category.
| Category | Early warning by email | Promotions and news by email | Licences | Incidents | Card issuer | Ratings | Methodology |
|---|---|---|---|---|---|---|---|
| Crypto Cards | yes | yes | yes | yes | yes | yes | Methodology |
| Crypto Proxies | no | yes | no | yes | no | yes | Methodology |
Early warning
When a provider loses its licence, halts withdrawals, gets hacked or a regulator warns about it, days can matter. In categories with early warning we email everyone who follows the product at once.
- Free and without an account, confirmed by email.
- Warnings go out at once, promotions and news at most once a day.
- We do not track whether you open or click an email. Unsubscribing takes one click.
- In categories without early warning, incidents still stay visible on the product page.
Errors and corrections
Despite all checks a value can be wrong or out of date. That is why every product has "Report an error".
- Every report becomes a review task. Corrections are based on the official source, not on the report alone.
- Every correction appears in the change history. We do not correct quietly.
- Readers, providers and we ourselves all go the same way.
What we are not
- Not a bank, not a payment provider, no investment, legal or tax advice. We do not broker contracts.
- Values can change before we check them again. The provider's current terms always prevail; the check date shows how current a value is.
- Crypto products can carry considerable risk, up to total loss.
Open to verification
Our data should be verifiable, not just our statements.
- All comparison data is under the free CC BY 4.0 licence, with a public API and a Markdown version of every page.
- We use no advertising trackers and no analytics cookies. Visits are counted anonymously.
- The network is operated by Connect Privacy LLC.