How trust works here

Every price-bearing card term — annual fee, APR type, secured-deposit range — carries its own source link and verification date. If a term isn’t reverified inside its hard-expiry window (45 days for annual fee and APR type, 60 for rewards, 30 for availability and welcome offers), the whole card is automatically dropped from the directory, sitemap, schema, and search — not flagged, removed. A card whose numbers might be stale is never shown as if they were current.

When we haven’t independently confirmed a fact about a lender — an enforcement record, a complaint count, a licensing detail — the site says so instead of assuming the best or worst case. There is no “assume clean” or “assume average” branch in the scoring code: a missing fact is left out of the weighted score entirely, and the exact share of the rubric a grade actually rests on is shown next to every letter grade.

A card or lender does not land on a warning list because an editor dislikes it. A card is flagged only when its own published APR or fee crosses a fixed percentile threshold computed from the current corpus, set before any card was measured against it. A lender’s negative marks come only from its own disclosed rate or a docketed public enforcement action, cited with a source URL and a retrieval date — never an adjective, never an inference. See the published lender-grading criteria.

Every partner in our affiliate registry ships inactive by default, pointing at the issuer’s own plain product page with no tracking and no commission, until a real network approval exists. More importantly, the code that computes a card’s rating or buyer-beware status has no field for “is this issuer a partner” at all — it cannot read that value even if it wanted to. See the advertiser disclosure for which relationships, if any, are active today.

We have zero consumer reviews published today, and we say so rather than seed the page to look busier than it is. Submissions that do come in are screened for Social Security numbers, likely full card numbers, and known scam phrasing before human moderation ever sees them, and any future average rating is withheld until at least five real reviews exist, so a tiny sample can never produce a misleading number. See how reviews work.

When we find, or are told about, an error, we correct the page and update its last-updated date. For material errors on pages carrying real legal or financial risk, we also note the correction on the page itself rather than silently rewriting history. Report one through the contact page — corrections are the most useful email we get.