MailFloo › How email verification works

How email verification works

Three steps, one question asked of the recipient's mail server, and a score built from its answer. With the actual SMTP transcript, because that is where the verdict comes from.

Updated 18 October 2026·9 min read

The question verification actually asks

Email verification does not send an email. It does what a sending server does in the first two seconds of delivery: it connects to the recipient's mail server and asks whether it would accept a message for that address, and then hangs up before any message is sent. The recipient's server answers with a three-digit code, and that code is the whole basis of the verdict.

Before any server is contacted, though, two cheaper rounds remove most of the obvious failures, so the slow network step runs on as few addresses as possible.

Step 1 · Local checks

These run in memory on the verifier's own machine, about a third of a second for a thousand addresses in MailFloo's case, and need no network at all.

  • Syntax. Is it structurally an address? One @, a valid local part, a domain with at least one dot. This catches pasted names, stray spaces and the classic john@gmail,com.
  • Typo correction. gmial.com, hotmal.com, outlok.com are flagged with the likely intended domain. MailFloo reports the suggestion rather than silently rewriting the address.
  • DNS and MX. Does the domain exist, and does it publish mail-exchanger records? A domain with no MX (or an MX pointing nowhere) cannot receive mail, so every address on it is invalid without any further work.
  • Disposable domains. A list of roughly 64,000 temporary-mailbox services (mailinator, guerrillamail and their thousands of rotating aliases), refreshed daily. These addresses often pass an SMTP check; they are only caught by knowing the domain.
  • Role and free-mail detection. info@, sales@, support@ are flagged as role addresses (a team, not a person); gmail.com, yahoo.com and the like as free providers. Neither is a verdict. Both are signals you may want to filter on.
  • Duplicates within the list are identified once and never charged twice.

Step 2 · A second opinion

Disposable and role detection is a moving target (new throwaway domains appear daily), so MailFloo cross-checks those two signals against an external service before the mail servers are contacted. Nothing here changes a valid/invalid verdict; it enriches the flags that travel with each address into your download and the API reply.

Step 3 · The SMTP conversation

This is the step that actually verifies. For each domain, the verifier connects to its mail server on port 25 and holds a scripted conversation:

→ EHLO mail.mailfloo.com
250 mx.acme.com Hello
→ MAIL FROM:<validator@mailfloo.com>
250 2.1.0 Sender ok
→ RCPT TO:<john.smith@acme.com>
250 2.1.5 Recipient ok          ← mailbox exists
→ RCPT TO:<zx9k7q-probe@acme.com>
550 5.1.1 User unknown           ← and the server is honest about unknowns
→ QUIT

The second RCPT TO is the important one. It names an address that cannot exist. If the server rejects it, the server is honest, and its 250 for the real address means the mailbox exists. If the server accepts the invented address too, the domain is catch-all, its replies are uninformative, and a different set of signals has to decide. MailFloo sends two invented probes, not one, so a single odd reply cannot mislabel a domain.

The reply codes the engine understands:

ReplyMeaningResult
250Recipient acceptedValid if the probes were rejected
550 5.1.1User unknownInvalid
550 5.4.1Microsoft 365: recipient not in directoryInvalid: a definitive no, not a server error
450 / 451Try again later (greylisting)Retried after 7 minutes, then once more
550 5.7.xPolicy rejection of the senderNot about the address. Logged as an identity problem on our side
No connectionFirewall, timeout, or the IP is blockedRetried against another MX; Unknown if nothing answers

Greylisting

Many servers reply 450 Try again later to any sender they have not seen before and only accept on the second attempt a few minutes later, a cheap anti-spam measure, since bulk senders rarely retry. A verifier that gives up on the first 450 reports those addresses as unknown. MailFloo waits the interval greylisting software actually requires (Postgrey's default is five minutes), asks again, and in the meantime tries the domain's backup mail server, which is often not greylisted at all.

Why your own setup matters

Mail servers judge the asker. If the verifier's IP has no reverse DNS, or its HELO name doesn't resolve, or there is no SPF record for the sending domain, many servers refuse the conversation outright, and a careless verifier reports that refusal as if the address were bad. MailFloo detects a refusal aimed at itself (550 5.7.1, "sender rejected", "spam") and reports it as an identity problem on the health page rather than marking your addresses invalid.

From replies to a score

A reply code is binary; confidence is not. Each address gets a 0–100 score that starts from the SMTP outcome and is adjusted by the signals around it: catch-all status, naming-pattern match, how the provider has behaved historically, whether the address is a role account, and (most strongly) whether a real message to it has ever been recorded as delivered or bounced. Bands are deliberately plain: 90+ very high, 70+ high, 40+ medium, under 40 low. And the bands are calibrated: as outcomes are recorded, MailFloo measures how often each band really delivered and scales future scores so that 80 keeps meaning roughly an 80% chance.

Where verification becomes learning

Most verifiers are stateless: the same address gives the same answer regardless of what happened the last time you sent to it. MailFloo keeps a loop open. A recorded delivery makes an address valid (score 97) even on a catch-all domain. A recorded bounce makes it invalid (score 8) even if the server accepted it. A domain that accepts and then bounces twice gets every future "valid" downgraded to risky. These updates reach the engine within ten minutes of being recorded, and outcomes can arrive by hand, by pasted list, by uploaded bounce report, or automatically from your sending tool's event webhook.

What happens to your data

Uploaded files and results are kept for 30 days so you can download them, then deleted. The evidence the engine learns from (which server said what, which domains accept everything) is kept in anonymised form, with each address replaced by an irreversible fingerprint. Nothing is shared with anyone, and no address on your list ever receives a message from the verification process.

Questions

Does email verification send an email to the address?

No. The verifier opens an SMTP conversation with the recipient's mail server, asks whether it would accept a message for the address, and closes the connection before any message is sent. The mailbox owner receives nothing.

How accurate is email verification?

For honest servers, an accepted or rejected address is close to certain. The uncertainty lives in catch-all domains, greylisting and servers that refuse to answer. A verifier's quality is largely how it handles those three cases; MailFloo resolves catch-all addresses with naming patterns and delivery history, retries greylisting after the interval it requires, and never reports a refusal of the verifier itself as a bad address.

What is greylisting in email verification?

A server replying 450 Try again later to a sender it has not seen before, and accepting only the second attempt a few minutes later. Verifiers that do not retry report those addresses as unknown.

What does an 'unknown' result mean?

The mail server would not give a usable answer even after retries: it timed out, refused the connection, or kept asking to try later. The address may be perfectly good; the server simply would not say.

Why does a valid email bounce anyway?

The server accepted the address at check time but the mailbox was full, the account was disabled between the check and the send, or the domain is catch-all and the address never existed. Recording the bounce in MailFloo turns that address invalid and tells the engine to be stricter with the domain.

How long does bulk verification take?

Local checks take under a second per thousand addresses. The SMTP step is the slow part and depends on how many distinct domains a list has; MailFloo runs domains in parallel and typically clears a few hundred addresses a minute per worker.