Editorial Guidelines

Our standards for trust and accuracy

Credit card decisions are financial decisions. These are the rules the site follows so that what it prints can be relied on — and so that what it cannot back is clearly marked.

Core principles

Say the number or say nothing

A figure prints only when it passes our evidence rules. One we cannot back is shown as “Not confirmed”, never as a zero or a guess.

Independence from issuers

The site has no affiliate partnerships and earns nothing from card issuers. Issuers do not see pages before they are published.

Rules, not opinions

Rankings are computed by the same published rules for every card. No card carries a hand-given score.

Fix errors in the open

When a figure or a sentence is wrong, we correct it under our public corrections policy.

How card figures are checked

Card pages, lists and rankings are generated from our catalogue by code, not written one by one, so there is no separate fact-check for each page. Instead, every figure passes the same rules before any page prints it:

Read from a document

Card figures come from documents issuers publish and from data regulators publish — the sources are listed on our methodology page. Some were extracted by automated tools, including an AI model.

Recorded with where it came from

For fees and interest rates, the catalogue records the document a figure was read from where there is one: its web address, when it was read, and the figure it stated.

Printed only under the evidence rules

A zero annual fee prints only where a check that read the fee from the issuer's terms confirmed it; a zero foreign transaction fee or interest rate only where a document states it; a welcome offer only where its text states a figure that has not expired. Anything else is shown as “Not confirmed”.

Dated only where a date is held

A card page dates a figure only where a check against the issuer's published terms is on record — “Checked” with the month, and a link to the document where we hold its address. A figure the issuer's pages we checked do not state is marked “Not published by the issuer where we checked” with the month we checked, and one behind a site that blocks automated reading is marked “Issuer restricts access — check their site”; neither is printed as a number. Every other figure is marked “Not yet checked”. The card's “Annual fee checked” date covers its annual fee and nothing else.

The full rules are on our methodology page.

Newsroom & AI-assistance policy

Our newsroom pairs automation with human oversight to publish accurate, timely rewards and consumer-finance news. This is exactly how it works, and how we disclose it.

Scope

We cover credit cards and card rewards, points and miles, airline and hotel loyalty programs, card issuers, consumer banking and fintech, and the regulators and central banks that affect them, across the seven markets we serve (Canada, US, UK, Australia, Brazil, India, and Portugal). We do not cover general news, politics, or topics outside consumer finance.

Sourcing — full-text official and newswire only

We summarize only from the full text of primary sources: official issuer, company, and regulator newsrooms, and established newswires (PR Newswire, Business Wire, GlobeNewswire). Aggregators are used strictly as a discovery signal to find a story — never as summarization input. If we cannot reach a full-text primary source, the story is not drafted or published. We never summarize from a snippet and never reconstruct a paywalled article. Facts are not copyrightable; wording is — so we write our own summary, attribute the outlet, and link back to the original.

AI-assistance disclosure

News summaries are drafted with AI assistance and published under our editorial-automation policy. Facts are sourced and verified against the primary source before publication. Every engine-produced article carries a visible disclosure line naming the source outlet and stating that the summary was drafted with AI assistance, alongside a “Report an error” link. The byline on every article is a real, named human — Riya Arora, Founder — who is accountable for the newsroom’s output. We do not invent authors, personas, headshots, or credentials.

Human oversight — a tiered model

Accuracy is graded before anything publishes. A story publishes automatically only when it clears a conservative accuracy gate: it must come from a trusted full-text source, be corroborated by independent reporting, pass our anti-hallucination verification (number-tracing, no verbatim copying, quote limits), score above a high confidence threshold, and not be a sensitive topic.

Everything else — anything uncertain, low-confidence, weakly corroborated, or sensitive — is held in a human review queue and published only after a named editor approves it. When we get something wrong, we fix it under our corrections policy, with a timestamp. Questions about our newsroom or ownership are answered on our About page.

Affiliate policy

PointsExplained has no affiliate partnerships and earns nothing from card issuers. Links to issuers go straight to the issuer's own website.

If that changes, any link that can earn a commission will be disclosed before it goes live — on our affiliate disclosure page and next to the link itself.

Corrections

Found a figure or a sentence that is wrong? Email corrections@pointsexplained.com with the page address and what you believe is wrong. How we check and label a correction is set out in our corrections policy.