Do not strip an international address as a bot

Answer Do not strip an internationalized address off a list because it looks strange in your spreadsheet. RFC 6531 defines the SMTP extension for internationalized email. An address with characters outside basic ASCII can be a real mailbox, when the path supports it. Unfamiliar is not the same word as fake.

The note that started this: RFC 6531, SMTP Extension for Internationalized Email, RFC 6531.

List vendors love a clean column. Clean, in a hurry, means delete anything the tool flags. The tool flags what the programmer found odd in the year it was written. Accented names, mailbox names in another script, and domains that are not a string of English words all land in the odd pile. Some of those are errors. Many of them are people. Safe solo ads sellers can tell the two apart without a superstition.

The specification calls the extension SMTPUTF8. When a server announces it, a client may send UTF-8 in the addresses and in header values that need it. When a server does not announce it, the client is not supposed to pretend. That is a capability negotiation. It is not a moral ranking of readers. A solo ad buyer who hears Unicode and thinks bot has confused a standards flag with a fraud score.

Can an email address contain characters outside basic English letters?

Yes. RFC 6531 defines an SMTP extension for internationalized email, so addresses and headers can use non-ASCII characters when both sides support it. Those addresses are addresses.

The reader with a name your keyboard does not type still has a mailbox. The mailbox may live at a provider that has supported these addresses for years. Treating the address as debris is a decision about your tooling, not a fact about the person. If your platform cannot submit the message to that address, the honest act is to say the platform cannot, and to skip it for that send. The dishonest act is to call the person a bot in the report so the delivered percentage looks tidy.

There is no census in this note of how many such addresses exist. Anyone who gives you a global percentage owes you a method. The specification does not include one. Your seller's export might include a count for their file. That count is their file. Ask for it as a count of addresses the platform refused, not as a count of frauds.

• A non-ASCII address is not auto-labeled as a bot.
• A platform limitation is labeled as a platform limitation.
• Removed addresses are counted separately from bounces.
• The buyer sees the rule before the send, not after.

Should a seller delete international addresses before a solo ad?

No. Unfamiliar characters are not proof of a bot. Deleting them by habit throws away real readers and does not prove the remaining file is clean.

The habit usually arrives from a deliverability myth. Someone had a bad week, noticed a cluster of addresses they could not read, and made a rule. The rule felt like hygiene. Hygiene would be a bounce, a complaint, or a confirmed typo. A script is not a bounce. A buyer who pays for a scrubbed file should ask what scrubbed means. If it means we deleted names we could not say out loud, you paid for a prejudice.

Real bots exist. They do not announce themselves with an accent. They announce themselves with behavior: addresses that bounce, addresses that never could have been typed by the person who supposedly joined, signups that arrived in a burst from one page. Chase the behavior. Leave the alphabet alone until the behavior shows up.

What should you ask a seller who scrubs the list?

Ask which rule removed the addresses, ask whether the sending path supports internationalized mail, and ask them not to call a character set a fraud signal.

Three questions. The rule should be something you can read in a sentence. The path should be a yes or a no on SMTPUTF8 from the platform, which is a fact their vendor can answer. The fraud label should be retired if the only evidence is the character set. A seller who needs the fraud label to justify a deletion is protecting a story.

Write the rule into the order if they insist on a scrub. You are buying a send to a defined file. A file that changes because a spreadsheet disliked a character is not the file in the quote. RFC 6531 is the reminder that the internet already has a name for these addresses. The name is addresses.