Stop an outreach sequence
The runtime switch is the fastest and the bluntest: it stops new work at once and needs nothing put back afterwards. Everything narrower — stopping the follow-up chain for one person or one thread, turning outreach autonomy down, recording a suppression — takes effect from the next piece of work rather than instantly, which is usually fine and occasionally not.
Ranked by how soon it bites#
| Action | Reach | Effective |
|---|---|---|
| Runtime switch | All new work, every channel | Immediately |
Set the channel's autonomy to off or draft_only | That channel, workspace-wide | From the next piece of work |
| Stop the chain for one person | Every follow-up for that person | From the next drain |
| Stop the chain for one thread | That conversation only | From the next drain |
| Record a suppression or do-not-contact entry | Outreach to that recipient, permanently | Checked before any outreach |
| Block the person | Every tagged channel, including future conversations | From the next piece of work |
Stopping one sequence properly#
If the wrong thing is going out to real people right now, use the runtime switch first and think afterwards.
Result New outreach stops. No autonomy mode, schedule or campaign configuration has changed, so nothing has to be reconstructed.
Find what is actually scheduled. Outreach continues through the follow-up queue, so the queue — not the prospect list — is where the remaining sends live.
Result You can see each pending step with its due date and its reason, which tells you whether this is one sequence or three.
Stop the chain at the right grain: for one person if the problem is who is being written to, for one thread if the problem is what the conversation became.
Result The remaining steps are cancelled rather than merely delayed. A snooze moves a send; it does not remove it.
Decide whether this is also a compliance event. If the person asked to stop hearing from you, record it as a suppression, an unsubscribe or a do-not-contact entry rather than as a cancelled follow-up.
Result Compliance is checked in one place before any outreach, so the record protects them from the next sequence as well as this one.
Switch the runtime back on and watch the follow-up queue for the period the next step would have fallen in.
Result Nothing new appears for that person or thread. That is the verification — an absence you deliberately looked for, at the time it would have shown.
Cold outreach and a customer reply are not the same path#
Outreach to a prospect and a reply to somebody who wrote to you run through different gates and are deliberately never merged. That matters here in both directions. Turning outreach off does not stop Connect answering a customer who emails in, which is almost always what you want. And stopping a sequence does not stop a conversation that the sequence started, once the prospect has replied — at that point it is a conversation, and it is governed by the channel's ordinary autonomy.
This is also why suppression is the durable answer for "never write to this person again" while a cancelled follow-up is the answer for "not this campaign". A cancelled follow-up leaves the person perfectly contactable by the next piece of work that finds them, and research discovery will find them again.
The ways this goes wrong#
- It stopped, then started again a week later
- The chain was stopped but the prospect was still qualified and contactable, so discovery produced them again. Suppression, or a block, is the durable stop.
- One person kept receiving messages
- They hold more than one identity — a second address, a number — and the stop was recorded against one of them. Merging the duplicates preserves identities from both sides and fixes it properly.
- Nothing is scheduled but a message went out anyway
- Look at whether it was outreach at all. A reply to an inbound message is a different path with different gates.
- The queue still shows the item
- A stopped follow-up and a completed one are different states. If it is still due, it was snoozed rather than stopped.
- You cannot tell what sent it
- The decision log names what was decided, by what, and under which rule — including refusals. Read it rather than inferring from the queue.
Questions#
Which stop is the one to reach for at two minutes' notice?
The runtime switch. It is the only control that needs no decision about scope, changes no configuration, and therefore leaves nothing to restore when you have worked out what the real fix is.
Does stopping a sequence delete the prospect research?
No. The prospect record, its evidence and its qualification survive; what stops is the scheduled contact. That is deliberate — the research was expensive and the decision not to write today is not a decision to forget the company.
If someone replies asking to be removed, is stopping the chain enough?
No. Record it on the compliance surface — an unsubscribe or a do-not-contact entry — so it is checked before any future outreach and carries its origin. A cancelled follow-up is not a record of their request.
Can the Assistant stop a sequence for me?
It can cancel and complete follow-ups, which covers stopping a chain. It cannot clear a do-not-contact entry — its rights are deliberately narrower than a person's on exactly this kind of decision.