An email verification result tells you how far a checker got before it stopped, not whether your email will arrive. Valid (or deliverable) means the recipient's mail server accepted the address in a live SMTP conversation; invalid means it was rejected or cannot receive mail at all; catch-all means the server accepts any address on that domain, so the mailbox could not be confirmed; unknown means the check never got a usable answer. Each label comes from a specific test, and each test has limits worth knowing before you decide what to send.
This guide is the hub for our email verification posts. It explains every check, every common verdict and what each one cannot prove, then gives a send, hold or skip table you can apply to a B2B list.
What a verifier actually checks
Every verifier, whatever it calls its labels, runs some version of the same pipeline. Results differ because tools stop at different points, ask from different places and interpret the answers differently.
| Step | What it tests | Typical failure label | What it cannot tell you |
|---|---|---|---|
| 1. Syntax | Is the address well formed under RFC 5322 and within SMTP length limits (64 octets for the local part, 255 for the domain, per RFC 5321 section 4.5.3.1) | Invalid (syntax) | Whether anyone owns it |
| 2. IDNA | Can an internationalised domain (for example café.example) be converted to its ASCII form | Invalid (domain) | Anything about the mailbox |
| 3. DNS | Does the domain exist at all | Invalid (no domain) or unknown (DNS timeout) | Whether it receives mail |
| 4. MX | Does the domain publish mail servers, a null MX, or nothing | Invalid (null MX) or domain valid | Whether the mailbox exists |
| 5. SMTP RCPT probe | Connect to the mail server and ask whether it will accept this recipient, then quit before sending | Invalid, valid, unknown | Whether the real message will be accepted or read |
| 6. Catch-all control | Ask the same server about a random address that cannot exist | Catch-all | Which addresses on that domain are real |
| 7. Lists and rules | Disposable domains, role addresses, known traps, previous bounces | Disposable, role, do not mail | Anything live; these are reference data |
Syntax and IDNA
Syntax checks catch typing errors and scraping debris: a missing @, two dots in a row, a trailing full stop, or a file name such as [email protected] that a careless extractor mistook for an address. They are fast and certain in one direction only: a syntax failure is definitely unusable, but a syntax pass proves nothing about the mailbox.
IDNA handling matters for UK lists more than people expect, because accented trading names occasionally appear in domains. A checker that cannot convert the domain to its ASCII form will report it as invalid when it is merely unusual.
DNS and MX
The checker looks up the domain's MX records. Three outcomes are possible:
- MX records exist. The domain names one or more mail servers. Good, but still says nothing about the mailbox.
- A null MX exists. RFC 7505 defines a single MX record with preference 0 and a target of
.as an explicit statement that the domain accepts no mail. A sending server should fail immediately with556 5.1.10rather than retry. Any address on such a domain is invalid. - No MX at all. Under RFC 5321 section 5.1, "If an empty list of MXs is returned, the address is treated as if it was associated with an implicit MX RR", meaning the sender falls back to the domain's own A or AAAA record. Many small business domains with web hosting and no email service look like this. A careful checker reports these as unconfirmed rather than invalid, then tries the implicit host.
A DNS timeout is not the same as a missing record. Good tools record "DNS unavailable" as unknown and retry, rather than calling the address invalid.
The SMTP RCPT probe
This is the only step that asks about the individual mailbox. The checker connects to the domain's mail server on port 25, introduces itself, gives a sender address, names the recipient with RCPT TO, reads the reply code and disconnects without sending a message. We walk through the full conversation in how to verify an email without sending.
The reply codes that matter are defined in RFC 5321 section 4.2.2:
| Reply to RCPT TO | RFC 5321 meaning | How verifiers read it |
|---|---|---|
250 | "Requested mail action okay, completed" | Accepted, pending the catch-all test |
251 | "User not local; will forward" | Accepted |
252 | "Cannot VRFY user, but will accept message and attempt delivery" | Accepted unconfirmed |
450, 451, 452 | Temporary failures | Unknown, retry later |
550 | "Requested action not taken: mailbox unavailable" | Invalid, if the text points at the mailbox |
553 | "Requested action not taken: mailbox name not allowed" | Invalid |
554 or 550 with policy text | Often a block on the checker, not the mailbox | Unknown |
The last row is where weak verifiers go wrong. A 550 that says 5.7.1 Service unavailable, client host blocked is about the checker's IP address, not the recipient. Reading the enhanced status code (5.1.1 means bad mailbox, 5.7.x means policy) and the reply text is what separates an honest result from a guess.
RFC 5321 also allows a server to accept the recipient now and decide later: some servers "do not perform recipient verification until after the message text is received" and treat a failure as a subsequent bounce. A 250 from those servers is weak evidence.
The catch-all control
After a 250 for the address you care about, a careful checker asks the same server about a random local part such as [email protected]. If that is accepted too, the server is accepting everything, and the earlier 250 tells you nothing about the real mailbox. The domain is labelled catch-all (some tools say accept-all).
Catch-all is a server setting. Organisations use it to avoid losing misaddressed mail, and many security gateways accept at the edge and filter later. We cover why large company domains so often look catch-all in catch-all on Microsoft 365 and Google Workspace.
Every verdict, what it means and what it cannot prove
Labels vary by vendor. This table maps the common ones to the evidence behind them.
| Verdict | Produced by | Meaning | What it cannot prove |
|---|---|---|---|
| Deliverable / valid | RCPT accepted, catch-all control rejected | The server knows this mailbox at the time of the check | That your message will be accepted, filtered to the inbox or read |
| Invalid | Syntax, null MX, no domain, or RCPT rejected with a mailbox code | The address cannot receive mail now | Nothing further needed; remove it |
| Catch-all / accept-all | RCPT accepted and catch-all control accepted | The domain accepts any local part | Whether this specific mailbox exists |
| Accepted unconfirmed | RCPT accepted but the catch-all control could not be completed, or the server replied 252 | The server said yes, but nothing confirmed that the yes was specific | Existence of the mailbox |
| Domain valid | DNS and MX checks only; no mailbox probe was run | The domain can receive mail | Anything about the mailbox |
| Unknown | Timeout, temporary error, greylisting, blocked connection, missing probe sender | The check did not finish | That the address is bad; this is missing evidence |
| Role address | Rule on the local part (info, sales, accounts) | A shared or functional inbox | Deliverability; role addresses are often valid |
| Disposable | Reference list of throwaway domains | A temporary address, rare in B2B data | Live status |
| Spam trap, abuse, do not mail | Vendor's own data | The vendor believes sending is risky | The vendor's method is usually not published |
Two vendor examples show how the same evidence gets different names. NeverBounce's result codes are valid, invalid, disposable, accept all (unverifiable) and unknown, and it notes that accept-all "is a setting on the recipient's mail server, not a verdict on the individual address". ZeroBounce's status codes add spamtrap, abuse and do_not_mail, and break unknown into sub-statuses such as greylisted, mail_server_temporary_error and timeout_exceeded. Those sub-statuses are useful because they tell you whether a retry is worth it.
Why results differ between verifiers
Run one list through three verifiers and you will get three different splits. That is expected, and the reasons are mostly mechanical.
- Different source IPs. Mail servers treat connections by reputation. A verifier whose IPs probe heavily may be rate limited or blocked, producing unknowns where another tool gets clean answers.
- Timing and greylisting. Greylisting servers reply
451to an unfamiliar sender and accept on a later attempt. A tool that retries after a few minutes gets an answer; one that does not reports unknown. See the SMTP cluster's greylisting guide for the mechanism. - Timeouts. RFC 5321 suggests a client wait 5 minutes for a reply to
RCPT. Verifiers working at speed give up after seconds, so slow servers become unknown. - Catch-all method. Some tools skip the control probe to save time and report a
250as valid. Others run it and report catch-all. The same domain gets two labels. - Edge filtering. Microsoft 365 tenants using Directory-Based Edge Blocking reject unknown recipients at the perimeter with
550 5.4.1 Recipient address rejected: Access denied, which gives a clear invalid. A domain behind a gateway that accepts everything gives catch-all from every tool. - Extra data. Vendors that add trap lists, complaint data or "inferred" results for catch-all domains are not reporting an SMTP answer at all. That can be useful, but it is a different kind of evidence and the method is rarely published.
- Date. Mailboxes close when people leave. A result from three months ago is three months old.
Decision table for B2B outreach
Verification answers "can this address receive mail". It does not answer "may I email this person", which is a legal question covered below. Assuming you have a lawful basis, here is a practical policy for cold outreach from your own mailboxes.
| Verdict | Where the address came from | Decision | Notes |
|---|---|---|---|
| Deliverable | Published on the company's site | Send | Still subject to permission and suppression checks |
| Deliverable | Guessed from a pattern | Send, normal volume | The probe confirmed what the guess could not |
| Catch-all | Published on the company's site | Send in a capped segment | Published addresses on catch-all domains usually exist |
| Catch-all | Guessed from a pattern | Hold or skip | Nothing has confirmed the mailbox |
| Accepted unconfirmed | Any | Treat as catch-all | Re-check if possible |
| Domain valid | Published | Hold until probed, or send in a tiny batch | No mailbox evidence yet |
| Unknown (timeout, greylist, temporary error) | Any | Hold and re-check later | Change network or time of day |
| Unknown (blocked, probe sender missing) | Any | Fix the setup, then re-check | Results reflect your checker, not the list |
| Invalid | Any | Skip and suppress | Do not re-check repeatedly |
Role address (info@, sales@) | Published | Usually send if the purpose fits | Check the inbox purpose label; skip privacy@, abuse@, noreply@ |
| Disposable, spam trap | Any | Skip | Rare in business data; investigate the source |
The biggest single improvement most senders can make is splitting catch-all and unknown into their own segments instead of lumping them with valid or deleting them outright.
A worked example
A fictional list of five addresses for a fictional joinery business, oakline-joinery.example, after verification:
| Address | Verdict | Evidence recorded | Decision |
|---|---|---|---|
[email protected] | Catch-all | RCPT 250; random control 250 | Published on contact page, so send in the capped segment |
[email protected] | Catch-all | Same server, same evidence | Guessed, so hold |
[email protected] | Catch-all | Same | Published in footer, purpose "sales", send in segment |
[email protected] | Catch-all | Same | Purpose "privacy", exclude from outreach |
[email protected] | Invalid | Domain does not exist | Typo in the source; fix or drop |
Notice that three addresses with identical verification evidence get three different decisions. Where the address came from and what the inbox is for matter as much as the verdict.
Verification is not permission
In the UK, PECR's rule on electronic marketing does not apply to corporate subscribers (limited companies, LLPs, public bodies), but you must still identify yourself and give a valid way to opt out, according to the ICO's business-to-business guidance. Sole traders and some partnerships count as individual subscribers, so you need consent or the soft opt-in. If an address identifies a person, such as j.patel@, UK GDPR applies: you need a lawful basis (often legitimate interests, after the three-part test), privacy information and respect for objections. Other countries differ. Germany generally requires prior consent for advertising email, including B2B, and the US CAN-SPAM Act sets its own rules. A deliverable verdict changes none of this.
How Map Extract helps
The email verifier in Map Extract runs the checks described above on your own PC and records the reason for every verdict rather than a bare label. It checks syntax and IDNA, looks up MX records with handling for null MX and implicit MX, and can run an optional SMTP mailbox probe on port 25 with a random-recipient catch-all control. Verdicts are deliverable, invalid, catch_all, accepted_unconfirmed, domain_valid (mailbox not checked) and unknown, with reasons such as dns_unavailable, mailbox_inconclusive or probe_sender_required. Role addresses are flagged and disposable domains are marked invalid. An optional Mailgun validation route is available if you have your own Mailgun account.
The email extractor keeps inbox purpose separate from verification, so privacy@ and noreply@ addresses are labelled and excluded from outreach even when they verify. In the bulk email sender, settings decide whether catch-all and unknown results may be sent at all, and the final send checks apply that policy again before each message goes out.
What it does not do: a verdict does not guarantee delivery or inbox placement, the SMTP probe needs an outbound port 25 connection that many home and office networks block (which produces unknowns), and the probe needs a sender address you control. It cannot turn a catch-all domain into a confirmed mailbox.
Related guides
- How to verify an email without sending: the SMTP conversation step by step.
- Catch-all on Microsoft 365 and Google Workspace: why enterprise domains look catch-all.
- Email list cleaner
For where verification sits in a full pre-send routine, see the cold email deliverability checklist.
Key takeaways
- A verdict reports how far the check got: syntax, DNS, MX, a live RCPT answer, or a catch-all control.
- Valid means accepted at the time of the check, not delivered, not read.
- Catch-all describes the domain's server, not the address; unknown is missing evidence, not bad evidence.
- Verifiers disagree because of IPs, timing, timeouts and method; read sub-statuses and reasons, not just the label.
- Decide with three inputs: verdict, where the address came from, and what the inbox is for.
- Verification never replaces a lawful basis, an opt-out and a suppression list.
Frequently asked questions
What does catch-all mean in email verification?
Catch-all means the receiving mail server accepted a made-up address on the same domain, so it would also accept the address you asked about whether or not that mailbox exists. The result describes the domain's server setting, not the individual address. Some catch-all addresses are real and read; others bounce later or land in a shared inbox nobody checks.
Is an unknown email verification result bad?
Not necessarily. Unknown means the check did not get a usable answer: the server timed out, deferred with a temporary error, greylisted the probe, or the verifier could not open a connection at all. It is an absence of evidence rather than evidence of a bad address. Re-check later, ideally from a different network, before deciding.
Why do two email verifiers give different results for the same address?
They ask from different IP addresses, at different times, with different timeouts and different rules for what counts as proof. A server may greylist one verifier, rate limit another, and answer a third. Some tools also add their own data, such as known spam traps or past bounces, and label the address on that basis rather than on a live SMTP answer.
Does a valid result guarantee the email will be delivered?
No. Valid means the recipient's server accepted the address during an SMTP conversation at the time of the check. The server can still reject the real message later because of content, sender reputation or authentication, or accept it and file it as spam. Mailboxes also close after the check. Treat valid as a strong signal with a date on it.
Should I send cold email to catch-all addresses?
Only in small, monitored batches, and only when the address has other support, such as being published on the company's own contact page. Guessed addresses on catch-all domains carry the most risk because nothing has confirmed them. Keep them in a separate segment, cap the volume and stop if bounces rise.
Sources
- ICO: Business-to-business marketing
- RFC 5321: Simple Mail Transfer Protocol
- RFC 5322: Internet Message Format
- RFC 7505: A "Null MX" No Service Resource Record
- NeverBounce help: Result codes
- ZeroBounce: Status codes (v2)
- Microsoft Learn: Use Directory-Based Edge Blocking to reject messages sent to invalid recipients
Written by the Map Extract editorial team. Found an error? Email [email protected] and we will check and correct it. How we research guides: editorial policy.