FACT CHECKING
Fact-Checking Policy
The unit of verification is the claim, not the article: a reviewer takes each factual statement and asks what it rests on, because a page can be well written, internally consistent and entirely wrong. Every claim traces to a source identified while the sentence is being written, not reconstructed afterwards. Sources are ranked and a claim is taken from the highest tier that covers it, so an interchange category comes from the card network rather than an article summarizing it. Unsourceable claims are removed, not hedged.
- 1. Regulators and government
- 2. Card network published rules
- 3. Domestic rail and debit scheme rules
- 4. A provider's own first-party sources
Try it, edit values and click buttons
API request
POST /v1/payments
{
"amount": 24900,
"currency": "USD",
"capture": true,
"metadata": { "order_id": "4821" }
}Response
Click "Send request" to see the API response
Processing payment…
201 Created
{ "id": "pay_7xK2m", "status": "approved", "amount": 24900 }Demonstration data only
The source hierarchy, applied in order
1. Regulators and government
Statute, regulation, regulator guidance and official statistics: the Federal Reserve, FinCEN, competition and privacy regulators and national statistical agencies. The highest tier because it is where the obligation itself originates rather than where someone described it.
2. Card network published rules
Visa, Mastercard and American Express published rules, interchange schedules and operating regulations, for anything concerning interchange categories, network assessments, chargeback rights, reason codes and acceptance requirements.
3. Domestic rail and debit scheme rules
The rules and specifications published by the domestic payments association for clearing, settlement and ACH debit, and by the domestic debit scheme — PIN debit worldwide — for debit acceptance. Used for any claim about how money moves between banks rather than between a merchant and a card network.
4. A provider's own first-party sources
For any claim about a specific provider — its pricing, its contract terms, its supported methods — that provider’s own published page, fee schedule or terms, recorded with the date the source was verified. Nothing outranks a company on the subject of what that company publishes about itself.
5. Standards bodies
The PCI Security Standards Council, EMVCo, the W3C, NIST and ISO for the standards they own. A requirement is quoted from the body that wrote it, not from a vendor page explaining what the body meant.
6. Peer-reviewed academic work
For economic or behavioural claims where published research exists, cited to the publication itself rather than to a summary of it, and read far enough to know what the paper actually measured.
7. Reputable industry publication
Last, and only for what no higher tier covers — a market development or an announcement not yet reflected in primary documents. Attributed by name, dated, and never used to support a fact one of the tiers above already answers.
Rules a claim has to survive
Sourced at the point of writing
The source is identified as the sentence is written. Sourcing a finished draft afterwards finds agreement rather than evidence, and the difference is invisible in the published text, which is exactly why the order matters.
The highest tier that covers it
Where a primary source exists, it is used. A secondary source summarizing a primary source is never cited in its place, however convenient the summary is and however often it has been repeated elsewhere.
Second-hand is labelled second-hand
Where only a secondary source exists, the page names it and says the claim is reported rather than established. A reader is entitled to know how far the evidence actually reaches before deciding how much weight to give it.
Numbers carry their denominators
A percentage publishes with what it is a percentage of. A price or rate publishes with the date it was verified. A figure with neither is a claim rather than a fact, whatever it is set in.
Claims about other companies come from those companies
Or from a regulator. A competitor’s marketing claim is not evidence about the market, and one provider’s blog is not evidence about a third provider, even where the two are describing the same event.
Unsupported claims are cut, not hedged
If it cannot be sourced it does not publish. Reportedly, widely and industry observers say are not sourcing; they are ways of keeping a sentence that failed review.
Related policies
Corrections policy
Where a disputed claim goes and what happens to it.
ExploreResearch methodology
The extra disclosures a number has to carry.
ExploreComparison methodology
First-party pricing evidence and verification dates in practice.
ExploreAI-assisted content policy
Why model output is verified more strictly, not less.
ExploreEditorial policy
Who checks the claims before a page publishes.
ExplorePayments glossary
Defined terms, sourced the same way as everything else.
Explore
Questions about this policy: Fact-Checking Policy
Why doesn't every claim carry a link?
Links are given wherever a public primary source exists. Some primary sources are not public — card network operating rules are licensed, and some standards documents are available only under agreement — and in those cases the page names the source and the section rather than linking to a secondary article that happens to be online.
What happens when two sources conflict?
The higher tier wins. Where two sources of the same tier conflict and neither resolves the other, the page states the conflict and what each source says, which is more useful to a reader than a confident sentence that picked one of them at random.
How is a claim about a competitor checked?
Against that competitor’s own published sources, with the verification date recorded on the page. Where they do not publish the detail, the page says it is not published rather than sourcing it from somewhere that guessed on their behalf.
Does this apply to AI-assisted drafts?
It applies more strictly. Model output reads as confident whether or not it is grounded, so every factual statement in an assisted draft is verified against a source before it can publish, and no statistic is ever accepted from a model. The AI-assisted content policy sets out the rest.
I think a source is being misread. How do I raise that?
Through the corrections policy: [email protected] with the page address, the sentence, and the source you believe it misstates. Misreading a source is a factual error and is handled as one, on the same targets as any other.
Take the next step
Talk to a RapidCents specialist
RapidCents Fee Check reads a processing statement and shows interchange separately from the markup. Upload a statement for an instant breakdown, or open a merchant account and start accepting payments on one account.
- No obligation
- Payment specialists, not a call centre
- Secure statement upload
