What is a catch-all email address?

Four ordinary reasons, none of them sinister. Typo tolerance, so mail to inof@ still reaches info@. Continuity, so mail addressed to someone who left is not lost. Legacy aliases nobody wants to audit. And a mail server default nobody ever changed.

It is a reasonable configuration with an inconvenient side effect, not a trick. The side effect is that the domain has given up the ability to tell anyone which of its addresses are real — including you.

Verification works by asking. The probe opens an SMTP conversation with the domain’s mail server, names a sender with MAIL FROM, names the address with RCPT TO, and reads the reply — then stops, before anything is sent.

On an ordinary domain that reply is informative. RCPT TO: <[email protected]> → 250 2.1.5 OK, and RCPT TO: <[email protected]> → 550 5.1.1 User unknown. The server distinguishes the two, so the answer is about the mailbox.

On a catch-all domain both replies are 250 2.1.5 OK. The absurd address gets the same acceptance as the real one, because the server accepts everything and sorts it out later. The 250 is now evidence about the domain, not about the mailbox — and no amount of protocol gets a different answer out of a server that has none.

That is also why retrying is pointless here. Unlike greylisting, this is not a transient condition; it is how the domain is configured, and it will answer the same way tomorrow.

For a discovered contact, a catch-all domain caps confidence rather than subtracting from it. A published address — one we can point at a page for — caps at 88. A generated address, produced by applying the domain’s pattern to a name, caps at 55.

They have to be ceilings. A percentage subtracted from a high raw score still clamps to a confident-looking number, so a penalty does nothing on exactly the records where it matters most. A ceiling cannot be outvoted by other positive signals.

Not delete them. In order: keep them, segmented separately from confirmed deliverables. Send at low volume first and watch what comes back. Prefer a published catch-all address over a generated one — the citation is the difference between evidence and arithmetic.

Never bulk-send a list that is mostly catch-all from a cold sending IP: you are combining an unmeasured list with an unestablished reputation, and the bounces arrive at the worst possible time. And do not schedule a re-verification hoping for a better answer. The answer is structural.

Ask the domain about an address that cannot exist. If a random local part is accepted, the domain is catch-all. The catch-all tester does exactly that and shows the reply it got.

The free tools answer three checks per IP per UTC day, across all of them, with no account. A free account is 100 credits a month, renewing, and a verification costs half a credit.