Unsubscribe

Acted on, not answered#

Triage classifies the message. unsubscribe sits in the small set of intents that receive no reply at all — alongside spam, autoresponders and internal chatter — because replying "noted, you have been removed" to somebody who has just asked to stop hearing from you is one more message they did not want. The absence of a reply is the correct behaviour here, not a gap.

Three things happen instead. The contact is marked do-not-contact. The conversation is closed, so it stops appearing as work waiting to be answered. And the scoring that would normally raise a warm contact's rank is suppressed for that message rather than applied.

Where the stop is enforced#

The mark is checked at the send boundary itself, before the mailbox is resolved and long before a provider is called. That placement is the whole point: there is one place an email leaves Connect, so there is one place the check has to be right. A refusal returns the reason — that this person asked not to be contacted — and is logged and audited like any other outcome.

They overlap on purpose. A person who unsubscribes from marketing has not necessarily asked the support desk to stop answering them, and a business that conflates the two either keeps mailing someone who opted out or stops replying to a live support case. The records are separate so the distinction survives.

The one-click header on promotional mail#

Cold outreach and marketing mail are sent through the provider's own list management, which attaches the standards-based unsubscribe header and the link that goes with it. Two constraints follow from that and are enforced rather than advised: exactly one recipient per promotional message, so the unsubscribe resolves to a real person rather than a list, and a contact list must be configured before such a message can be sent at all.

If the text or HTML body does not already carry the provider's unsubscribe token, one is appended before the message goes. A promotional message without a working way out is not sent in a reduced form; it is not sent.

Before a live promotional send, the recipient's subscription state is read back from the provider. An address that has opted out of everything, or out of the topic being sent, is refused. Where that state cannot be verified the send fails closed rather than proceeding on an assumption — the point of checking is to make an unsubscribe irreversible, and a check that gives up when it is inconvenient is not a check.

What it does not do#

It does not stop their mail arriving. Inbound messages still sync, still thread and still appear in the conversation list — an unsubscribe governs what leaves, not what comes in, and the message that later says "actually, please do get in touch" has to be able to reach you.

It does not delete the relationship. The person, the history and the timeline stay; what changes is that nothing is sent to them. Removing the record entirely is a different request, answered under Security and isolation.

And it cannot be undone by the Assistant. Clearing a do-not-contact entry is outside the Assistant's rights on purpose — its authority is deliberately narrower than a person's, and a compliance record is exactly the kind of thing that should need a human hand.

Questions#

Related#

• Do not contact
• The suppression list
• Outreach email versus a customer reply
• Spam complaints
• Connect Assistant