Methodology

Core principles

• Real-world reports only — no simulated or estimated data
• No guessing — a low-confidence match is treated as no match
• Unknown is better than wrong — a blank field beats an incorrect one
• Test dates are always shown, and stale data is marked stale

Where institution data comes from

Bank and credit union contact info (website, address, phone) comes from official regulatory sources, not self-reported by users:

• Banks — FDIC BankFind (website, address)
• Credit unions — NCUA's quarterly call report data (website, address, phone); no live API exists for this, so it's synced periodically from NCUA's bulk data files
• Brokerages — FINRA BrokerCheck (address, phone only — no official website field exists for broker-dealers in any regulatory source we checked)

Matching a user-entered institution name against these sources never guesses. If a name is ambiguous enough to plausibly match more than one real institution, it's treated as no match rather than picking one — a missing field is always safer than a wrong one.

InstantRailCheck is an independent service. References to FDIC, NCUA, the Federal Reserve, The Clearing House, Zelle, and listed financial institutions identify data sources or payment networks only; they do not imply affiliation, endorsement, sponsorship, or approval.

Institution names, aliases, and source attribution

Which regulator an institution's contact info is sourced from (shown on its profile page) is decided by that institution's actual regulatory record — never guessed from wording in its name. A credit union whose name doesn't happen to contain the words "credit union" is still correctly labeled as NCUA-sourced.

Regulators' own raw source files sometimes include a self-registered trade name or alternate name for an institution. That data isn't automatically shown as a public "also known as" alias just because it appears in an official file — it's checked first for a genuine relationship to the institution's real name. Alternate names that instead name an unrelated, unaffiliated company are suppressed from public display, even though the raw source data remains available for internal audit and search purposes. An institution's own legitimate abbreviation or acronym (verified independently, not assumed) can still be shown.

Payment rail verification

Each rail is verified differently, based on what's actually available:

• FedNow — checked against the Federal Reserve's official participant list
• RTP — checked against The Clearing House's official participant list
• P2P - Zelle — checked against Zelle's own partner directory, which is confirmed to be incomplete (a genuine Zelle-supporting bank can be absent from it). A missing badge means "not listed," not "confirmed unsupported."
• ACH, Wire, Visa Direct, Mastercard Send — no accessible official directory exists for any of these, so they're tracked entirely from user-submitted route reports

Name-matching, precisely

Verifying rail participation means matching an institution's name against an official participant list — harder than it sounds, since official lists use full legal names ("Capital One, National Association") while most people think in brand names ("Capital One"). The matcher tries progressively shorter prefixes of a name, but only trusts a substring match on the complete, untruncated name — and only if it resolves to exactly one distinct institution.

That last part matters: a generic word like "Farmers" legitimately appears in two dozen unrelated "Farmers ... Bank" entities across official lists. Matching on it would produce a confident-looking but essentially random result. Multiple matches are treated as ambiguous — not a guess, no match.

Route evidence

Every searched route shows evidence rather than a single numeric confidence score. Only reports from a signed-in account count, and only each reporter's newest report for a given directional route (sending bank → receiving bank) and rail — submitting the same report repeatedly can't inflate it. The reverse direction (receiving bank → sending bank) is tracked as a completely separate route, since rail support and timing aren't guaranteed to be symmetric.

Only reports from the last 180 days count toward an active evidence state. If every report for a route is older than that, it's shown as Previously observed rather than disappearing — the evidence still exists, it's just no longer treated as current.

A route's evidence state is one of:

• Limited evidence — exactly one reporter within the last 180 days
• Observed working — two reporters within 180 days, all successful
• Consistently reported — three or more reporters within 180 days, all successful
• Variable timing — a mix of successful and delayed reports, no failures
• Reported delayed — every recent report was delayed
• Reported unsuccessful — every recent report failed
• Conflicting reports — recent reports disagree (at least one failure alongside a success or delay)
• Previously observed — evidence exists, but all of it is older than 180 days

These states describe what's been observed and reported — not a guarantee of what will happen on your own transfer.

Early Direct Deposit leaderboard

The Early Direct Deposit leaderboard ranks institutions by how early community members have reported their direct deposits arriving. Only each reporter's newest report per institution counts, and inactive institutions never appear regardless of how much historical evidence they have.

A ranked position requires at least 5 distinct reporters for that institution; 2-4 reporters appear only in an unranked "early evidence" section, not the ranked leaderboard. The headline number is the typical (median) reported timing, not a raw average — a single report of "more than 5 days early" is a real observation, not an unbounded exact count, so it's never averaged in as though it meant literally six days. When the typical value would need to reflect one of those open-ended reports, it's shown in plain language ("more than 5 days early") instead of a fabricated number.

Each ranked institution shows a sample-size label reflecting how much evidence backs it — not certainty: Emerging evidence (5-9 reporters), Moderate evidence (10-24), Strong evidence (25+) — plus the date of the most recent qualifying report, marked stale if none has come in within 180 days.

Direct deposit timing ultimately depends on when an employer, benefits agency, or payroll provider sends the payment file — a bank's past timing doesn't guarantee your own deposit will arrive at the reported time.

Settlement-time leaderboard

The settlement-time leaderboard ranks institutions separately by rail and by which side sent the transfer, using the typical (median) settlement time rather than a raw average, for the same reason as Early Direct Deposit: a single unusually slow (or fast) report shouldn't be able to swing the headline number the way it would swing an average.

Failed transfers and invalid negative values are excluded entirely — they don't have a meaningful settlement time. A delayed-but-eventually-completed transfer still counts, since it did settle, just later than expected. As with route evidence, only each reporter's newest report for a given directional route and rail counts, at least 2 distinct reporters are required for an institution to appear, and inactive institutions are excluded. Ties are broken deterministically (fastest typical time, then consistency, then sample size, then name), and the same sample-size evidence labels and 180-day staleness marking used for Early Direct Deposit apply here too.

Corrections never override confirmed data

Automated re-syncs against official sources never downgrade a rail flag that's already confirmed true — a positive confirmation (even a manual correction) always outweighs a later absence in a source that can itself be incomplete, like Zelle's directory. User-suggested corrections to a bank's website or phone are re-verified against the same official sources before being applied; a match auto-applies, a mismatch is flagged for review instead of trusted blindly.