PRODUCT BUILD — CASE STUDY

The verifier that asks the mail server, not the format

EmailValidatorTool, our own product

3 secondsfor a real SMTP verdict on a fake Gmail address, answered with Gmail's own 550 code
~3 secfor a real SMTP verdict on a single address
38rows in the live bulk test, all done in under 45 seconds
10,000rows in the splitter test, split into 7 parts, none lost
40recipients checked per SMTP connection
6builds shipped on launch day alone, up to v2.10
ST7212029the Spamhaus delisting case, filed on launch day

How it worked

STEP 1

Real SMTP handshake, RCPT TO asked per address

STEP 2

Random-address catch-all probe per domain

STEP 3

MX, syntax, disposable and role-based checks

STEP 4

Four honest statuses, valid, invalid, risky, unknown

OUTCOME

3 seconds for a real SMTP verdict on a fake Gmail address, answered with Gmail's own 550 code

Meet the product

EmailValidatorTool is our own email verification service, built because list quality is where every campaign in this case study series either starts healthy or starts dying.

The challenge

Most verification is theatre: check the format, check that a mail server exists, declare the address valid. On catch-all domains, where the server accepts everything, that method calls dead addresses valid, and the sender pays for the lie in bounces. We know this failure mode first-hand: a shallower checker of our own had exactly this weakness, and we took it offline rather than sell guesses.

The solution

Ask the server, per domain

The verifier performs the real SMTP handshake and asks about the actual address with RCPT TO, then reads the server's own answer. A fake Gmail address came back invalid in about three seconds, carrying Gmail's real 550 5.1.1 response. That is a verdict from the mail server, not an inference. It checks up to 40 recipients per connection, with a 2 second pace between connections and parallel domain workers, so it asks firmly without hammering the servers it talks to.

Probe for catch-alls

For each domain it also asks about a random address that cannot exist. If the server accepts that too, the domain is catch-all and its results are marked honestly instead of being dressed up as valid.

Statuses that admit doubt

Every address lands in one of four states: valid, invalid, risky or unknown, with a fifth, unchecked, used by the lite mode when outbound port 25 is blocked. Disposable and role-based addresses are flagged. Big accept-all providers are short-circuited with rules drawn from their real behaviour. Results are cached per domain.

The results

Live on September 25, 2026, from build to public HTTPS the same day, and iterated hard: six builds shipped on launch day alone, up to v2.10, which the live site still runs.

The proof numbers from live testing: a 40 line bulk CSV came back with 38 rows accepted (2 malformed rows rejected at input) and all 38 finished in under 45 seconds, landing at 16 invalid, 12 risky and 10 unknown. A separate splitter test pushed 10,000 rows through the CSV splitter and produced 7 parts with all 10,000 rows preserved.

And the limit, stated plainly: verification of Microsoft addresses is blocked because the server's IP sits on a Spamhaus list, and Microsoft refuses the conversation at the door. A delisting case, ST7212029, was filed the same day the tool launched. Gmail, Yahoo and corporate domains verify normally. A verifier that hides its blind spots is not a verifier, so ours are on this page.

The return shows up in what does not happen. In the live bulk test, 16 invalid and 12 risky addresses were caught before sending, at about three seconds per address. Sent blind, every one of those is a bounce or a spam signal charged against the sender. Catching them first is the cheapest deliverability insurance there is.

FAQ

Why does an SMTP handshake beat a format check?

Because it asks the only authority that knows: the receiving mail server. Format and MX checks can only say an address could exist. The handshake gets the server's actual answer about the actual mailbox.

What does risky mean?

Usually a catch-all domain, a role-based address, or a disposable address: reachable, perhaps, but with a real chance of bouncing or never being read by a person. Risky is a warning label, not a verdict.

Why can Microsoft domains fail?

Microsoft blocks the opening of the SMTP conversation from listed IPs before any address is asked about. It is a server reputation issue on our side, disclosed openly, with a delisting case in progress.

By the numbers

for a real SMTP verdict on a single address~3 sec
rows in the live bulk test, all done in under 45 seconds38
rows in the splitter test, split into 7 parts, none lost10,000
recipients checked per SMTP connection40
builds shipped on launch day alone, up to v2.106
the Spamhaus delisting case, filed on launch dayST7212029

Bars are scaled so very small and very large numbers fit on one chart. The values printed are exact.

THE ROI
28 of 38

addresses in the live bulk test flagged invalid or risky before a single email was sent to them

In the live bulk test, 28 of 38 addresses were flagged before sending: bounces that never hit a sender.

What was built

The concrete pieces this engagement shipped, in one list.

Your mail is being sent unauthenticated. Here is the full fix.

Lower stack cost. Back in the inbox. Built to send at any scale.

We rebuild it properly: secondary domains, inboxes, SPF, DKIM and DMARC set to enforcement, warmup and safe sending limits, blacklist monitoring on top. Send-ready in about 17 days. No campaigns, no copy, no lead lists. Infrastructure only.

Almost a decade building cold outreach infrastructure, including his own sending and verification stack built from scratch. 20+ B2B clients at companies worth $10M and above, all in tier one countries.

Free · Written plan + fixed price · No sales call required

More case studies

Lead dataA US M&A advisory firm (name withheld)
46,000

prospects matched on LinkedIn, 93% of the final list

151,218 rows in. 49,373 decision-makers out.

READ CASE STUDY →

ROIValid decision-maker leads at $0.067 each in the pilot, then all 49,373 scored for $1.54 total.

InfrastructureA US advisory firm's cold email program (name withheld)
100%

sender health at the end of the wave, with 1 hard bounce in the final campaign

30 inboxes. 1,596 sends. Zero burned.

READ CASE STUDY →

ROIZero inboxes burned: 1,596 sends delivered and the pool ended at 100% health, ready for the next wave.

OutreachPartner outreach for a US advisory firm (name withheld)
Step 2

the follow-up, not the first touch, produced the first interested partner reply

The first yes did not come from the first email

READ CASE STUDY →

ROI958 sends across the wave produced an interested partner and a $1,000 logged opportunity.

Copied
Free tools Get an inbox plan