Most campaign damage happens in the hours nobody is looking. A message that lands badly at 09:00 keeps landing badly until someone opens the dashboard at 14:00, and by then five hundred people have received it and a percentage of them have pressed Report.
The fix is not more monitoring. It is deciding, in advance, what would make you stop, and letting the campaign enforce that on its own. A stop rule written before launch is a policy. The same decision made at 14:00 under pressure is a negotiation, and it usually ends with “let it finish”.
The six rules worth having
1. Stop on reply
Non-negotiable, and the one most tools get right. Nobody receives step two after answering step one. It is basic courtesy, it prevents the single most irritating outreach experience there is, and it removes your most likely reporters from the sequence.
The subtle part is what counts as a reply. A “no thanks” is a reply. So is an emoji. Anything from them means the sequence is over and a human takes it from here.
2. Stop on block or report signal
If people are blocking you, the message is the problem and more sends will not fix it. Set the threshold low, because by the time this number is large the damage is done. One block in the first twenty recipients is worth a pause and a read of the copy.
3. Stop on send failures
Errors are the earliest signal your infrastructure gets. If more than a small fraction of sends fail in a window, pause the whole run rather than retrying. Retrying into a wall converts a temporary problem into a permanent one.
A specific case deserves its own hair trigger: the first PEER_FLOOD stops the campaign for that account, full stop. Not “reduce the rate”. Stop, and leave the account alone for a few days. More on what those errors mean in Telegram DM limits.
4. Stop on a dead reply rate
This is the quality rule, and it is the one people leave out. Decide a floor before launch: if after the first hundred sends the reply rate is under some percentage you named in advance, the campaign stops for review.
The number matters less than having named it. Teams that skip this run campaigns to completion that were visibly failing after twenty messages, and pay for it in reputation rather than money.
5. Stop outside working hours
Nothing sends between 21:00 and 08:00 in the recipient’s timezone. A 03:00 sales message is a report waiting to happen, and it says something about you that no amount of good copy recovers from.
Weekends are a judgement call by market. Most B2B markets should not be receiving outreach on Sunday.
6. Stop at the cap
A daily ceiling per account, and a total ceiling per campaign, both set before launch. The purpose of a cap is not safety in itself. It is that it forces a human to look at the results before deciding to send more, which is exactly the checkpoint that gets skipped otherwise.
Thresholds we would actually use
Numbers are context-dependent, but empty advice helps nobody, so here is a starting configuration for a warm campaign from an established account.
- Reply rate floor: review below 5% after the first 100 sends.
- Block signal: pause at 1% of sends, or any 2 blocks in the first 50.
- Error rate: pause at 3% of sends failing within an hour.
- PEER_FLOOD: stop immediately, first occurrence, that account only.
- Daily cap per account: low enough that a bad day is survivable. Start conservatively and raise it only while reply rate holds.
- Quiet hours: 21:00 to 08:00, recipient local time.
Adjust from your own data after three campaigns. Do not adjust upward because a campaign is performing well on day one.
Restart is a separate decision
A stopped campaign should not resume on its own. Someone has to look at why it stopped and answer three questions: what triggered it, what changed as a result, and what threshold applies now.
The most common mistake here is restarting with the same message to the remaining list because “the first batch was probably just a bad segment”. Sometimes true. Usually the message is the problem, and the remaining list is about to prove it more expensively.
If you do restart, restart smaller. A hundred more sends of a message you have revised tells you more than the rest of the list at once, and it costs less if you are wrong.
What a stop rule cannot save you from
Worth being clear: none of this makes a cold blast to a scraped list safe. Stop rules limit the damage of a campaign that turns out to be wrong. They do not make a campaign right.
The things that decide whether outreach works are upstream: whether the recipient has any reason to know you, whether the message is about them, and whether the offer makes sense at that moment. Stop rules are the seatbelt. They are not the driving.
Where Cavyro fits
Campaigns in Cavyro are built for warm outreach through your own accounts, with pacing, per-account daily caps, quiet hours, stop-on-reply and automatic pauses on Telegram warning signals. Every send is attributed, so the reply rate you are stopping on is a real number rather than an estimate, and revenue per campaign shows up in reports rather than in a separate sheet.
If your current process is “send it and check later”, start a free 5-day trial and run one campaign with the rules on. The difference is mostly in the nights.