Bulk Email Verifier — validate and clean an email list before you send

Give it a list of email addresses. Get back a deliverability verdict for every one — as a clean table you can export to CSV, Excel or JSON, or push straight into your sending tool.

It exists because a scraped list is never a send list. Somewhere between 10% and 30% of the addresses on one are dead domains, typos, shared inboxes or throwaway mailboxes, and every one of them is a bounce. Enough bounces and your domain reputation goes with them.

---

What it checks

Each address goes through five layers, cheapest first, and stops as soon as there is a definite answer.

Layer	Catches
Syntax	Malformed addresses, illegal characters, over-length local parts, missing or bogus TLDs
Typo detection	sid@gmial.com → did you mean sid@gmail.com — transposition-aware, so the most common real typo is caught
Domain	Domains that do not exist, and domains that publish a null MX (RFC 7505 — "this domain never receives mail")
Mail server	MX records, with the RFC 5321 fallback to an A record, so small domains that really do receive mail are not thrown away
Mailbox (optional)	An SMTP conversation as far as RCPT TO — no message is ever sent — plus catch-all detection so a meaningless 250 is not read as a real mailbox

Things most verifiers get wrong, and this one handles

• Catch-all servers. A domain that accepts every address will happily accept a mailbox that does not exist. We probe a random address first, and if it is accepted the whole domain is marked isCatchAll and scored as risky rather than sold to you as deliverable.
• Gmail normalization. J.Smith+news@gmail.com and jsmith@gmail.com are one mailbox. The normalizedEmail column collapses them, so you can deduplicate a list properly instead of paying to mail the same person twice.
• Role vs personal. info@ and sales@ reach a shared inbox. They are still valid, so they are still deliverable — but they get their own isRoleAccount column, because cold outreach usually excludes them and a silent drop would be the wrong call to make on your behalf.
• Public suffixes are not disposable. Auto-generated blocklists tend to swallow co.uk and edu.pl because throwaway services hand out subdomains on them. Flagging every UK business as disposable is a far more expensive mistake than missing a throwaway, so the parent-domain list here is curated by hand.
• Honest unknown. Where the mailbox could not be proven, the status is unknown — not "probably fine". See the note on port 25 below.
• The mail platform. Free by-product of the MX lookup: Google Workspace, Microsoft 365, Zoho, Proofpoint, Mimecast and 25 others, identified per domain.

---

Input

Field	Type	Default	What it does
emails	array	—	The addresses. Paste one per line, upload a CSV, or link another Actor's dataset. Display names and mailto: prefixes are handled.
smtpCheck	boolean	true	Probe the mailbox over SMTP. See the note below.
onlyKeepSendable	boolean	false	Drop undeliverable rows so the export is already a clean send list.
maxEmails	integer	0	Safety cap while testing. 0 = no limit.
maxConcurrency	integer	20	Addresses handled at once.
dnsTimeout	integer	5	Seconds to wait for an MX record.
smtpTimeout	integer	8	Seconds to wait on the mail server.
heloDomain	string	example.com	The domain the probe introduces itself with. Set it to one you own for large lists — servers trust a real domain more.
mailFrom	string	verify@example.com	Return address used during the probe. Nothing is ever sent to it.

Minimal input:

{
  "emails": [
    "info@apify.com",
    "sid@gmial.com",
    "test@mailinator.com",
    "nobody@nosuchdomain-zzq7.com"
  ]
}

---

Output

One row per address.

{
  "email": "sid@gmial.com",
  "normalizedEmail": "sid@gmial.com",
  "status": "risky",
  "score": 30,
  "reason": "likely_typo",
  "localPart": "sid",
  "domain": "gmial.com",
  "isRoleAccount": false,
  "isFreeProvider": false,
  "isDisposable": true,
  "isCatchAll": false,
  "hasMxRecord": true,
  "mxProvider": "Other",
  "mxHost": "mx.gmial.com",
  "didYouMean": "sid@gmail.com",
  "checkLevel": "dns",
  "smtpDetail": ""
}

The status column

Four values, and they are the one thing in this Actor that is never fudged.

Status	Means	What to do
deliverable	The mailbox exists and accepted the address	Send
risky	Real domain, but something is wrong with it — catch-all, disposable, or a likely typo	Send at your own risk, or fix the typo first
undeliverable	Guaranteed bounce — bad syntax, dead domain, null MX, or the server rejected the mailbox	Remove
unknown	The domain is fine but the mailbox could not be proven	Your call — score 65 means everything short of the mailbox check passed

score is 0–100 and is there for sorting; reason is a stable machine-readable code you can filter on.

---

A straight note about port 25 and unknown

The mailbox probe needs outbound port 25, and many networks block it. When that happens the Actor:

1. notices after a handful of domains,
2. stops probing rather than burning your compute on timeouts that all say the same thing, and
3. returns unknown with reason: mailbox_unverified_mx_valid for the rest.

A DNS-level run is still worth running. It removes bad syntax, dead domains, null-MX domains, typos and disposable addresses — which on a scraped list is the large majority of what would have bounced. What it cannot tell you is whether one specific mailbox at a live domain exists. The Actor says so instead of guessing, because a verifier that quietly upgrades "I could not check" to "deliverable" is worse than no verifier at all.

---

Where this sits in a pipeline

Google Maps / directory scrape   →   company names + websites
            ↓
Website Contact Extractor        →   emails, phones, social profiles
            ↓
Bulk Email Verifier  (you are here)  →   a list that is safe to send to

Link the dataset from Website Contact Extractor directly into the emails field — its emails and emailsRole columns are read automatically, so no reshaping is needed in between.

---

Pricing

$1.00 per 1,000 emails ($0.001 each). No monthly fee, no minimum, nothing to cancel — you pay for the addresses you actually put through it, and a failed check costs the same as a successful one because the work happens either way.

For comparison, the mainstream web services charge $3–$16 per 1,000 for the same job.

---

Run it locally first

No Apify account needed:

$python tools/try_local.py info@apify.com sid@gmial.com test@mailinator.com

Add --no-smtp for DNS-level checks only, or --file list.txt to read a file.

Unit tests, no network required:

$python tests/test_verify.py

---

Notes

• The disposable-domain list is bundled from martenson/disposable-email-domains (CC0). Refresh it with python tools/refresh_disposable.py.
• No message is ever sent to any address you submit. The SMTP conversation stops at RCPT TO and issues QUIT.
• Addresses you submit are processed for the run and written only to your own dataset.