Drip campaigns that stop when someone replies
An email sequence that keeps going after somebody answers is rude. A WhatsApp sequence that does it is a report, and reports are what bans are made of. The rule is simple and almost nobody implements it properly.
Email sequences taught a generation of marketers a habit that is actively dangerous on WhatsApp: send message two whether or not anybody answered message one.
In email that is mildly rude and costs an unsubscribe. On WhatsApp it is a person being talked over in a one-to-one conversation, and the button they reach for is Block, which is the strongest negative signal the platform has.
The one rule
The moment somebody replies, the sequence stops. Not pauses. Stops. A reply means this is a conversation now, and a scheduled message arriving on top of a conversation is the clearest possible evidence that nobody is reading.
Almost every tool claims to do this. A useful number of them implement a version that breaks in one of the ways below.
Four ways “stops on reply” fails
| The bug | What the customer sees |
|---|---|
| Stops only the current step | They answer, get a polite pause, and message three arrives on Thursday anyway |
| Checks at send time, not on arrival | They replied four minutes ago; the message was already queued and goes out regardless |
| Only counts replies in the same thread | They phoned you instead, or answered a different number of yours, and the drip continues |
| Enrols them again from another trigger | They finish one sequence and a second starts, which from their side is the same thing |
The last is the most common in mature setups: several flows, each individually polite, collectively producing a stream of unprompted messages nobody meant to send.
What a sequence should look like
Shorter than an email sequence, by a lot.
- Two or three messages, maximum. Not seven. If three did not land, the answer is no.
- Spaced in days, not hours. Same-day follow-ups read as pressure.
- Each one useful alone. If message two is meaningless without message one, it will be meaningless, because plenty of people did not read message one.
- Varied wording. Identical text at scale is trivially detectable and looks exactly like what it is.
- An obvious way to stop it. “Reply STOP” costs you almost nobody and is the difference between an opt-out and a report.
Timing and pacing
Two separate concerns, and both get numbers wrong in the same direction.
When a person receives it
Quiet hours are not optional. A marketing message at 11pm is a message from a business that does not respect the reader, and the reply rate reflects it. Their local time, not yours.
How fast the number sends
Two hundred identical messages in ten minutes is the most recognisable spam signature there is. Real conversations are paced and irregular. A sequence going to a thousand people should trickle across hours, and a new number should send far less than an established one — see what gets numbers banned.
Sequences that are actually welcome
| Sequence | Why it works |
|---|---|
| Post-purchase: confirm, ship, arrive | Expected, and about something they bought |
| Onboarding after signup | They just asked for this |
| Appointment: booked, reminder day before | Genuinely useful, high reply rate |
| One abandoned-cart nudge | Works if it is one — see the tutorial |
| Re-engagement after months of silence | Fine once. A sequence of them is not |
| Cold outreach to a purchased list | Do not. This is how numbers die |
Measuring the right thing
Open rates do not exist here and delivery rates are near-universal, which leaves people measuring sends. The numbers that matter are the uncomfortable ones:
- Reply rate per step. Falling sharply by step two means the sequence is too long.
- Block rate. The single number most predictive of losing the number.
- Stop requests. Rising means the targeting is wrong, not the copy.
- Delivered versus accepted. A widening gap is an early warning — see sent but not delivered.
In WaveSenda
A flow follows one contact and stops when they reply — across the flow, not the step, and checked when the message is about to go out rather than when it was scheduled. Quiet hours and per-number pacing apply to every send, and a number that starts attracting blocks has its sending slowed rather than continuing into a ban.
For a one-off announcement to a list, that is a campaign rather than a flow — the distinction matters, because a campaign has no conversation to interrupt.
Common questions
How many messages should a WhatsApp sequence have?
Two or three. Email norms of five to seven do not transfer: each message is a one-to-one interruption on a personal device, and the marginal message past three converts very little while adding real block risk.
What counts as a reply that should stop a sequence?
Any inbound message from that contact, including a short one. "Ok" is a reply. Someone who has engaged at all is in a conversation, and a scheduled message landing on top of it undoes the engagement.
Can I send WhatsApp marketing to a purchased list?
No, in the practical sense that it will cost you the number. Messaging people who never contacted you is the single largest cause of restrictions, and a purchased list is that by definition.
What is the difference between a flow and a campaign?
A flow follows one contact over time and reacts to what they do — it stops when they reply. A campaign is one message to many people at once. Using a flow for a broadcast makes it slow, and a campaign for a conversation makes it deaf.
Do I need consent to send a sequence?
Practically, yes, whatever your local law says. The enforcement mechanism is recipients pressing Block and Report, and people who did not ask to hear from you press them at a far higher rate. Consent is a deliverability strategy before it is a legal one.