Automating follow-ups without becoming a nuisance
Most enquiries do not evaporate because there is no interest, but because nobody follows up after the first attempt. Four attempts at widening intervals are defensible — provided each one has a reason of its own.
In short
- Four attempts at widening intervals are defensible — day 3, day 8, day 18, day 40.
- The difference between helpful and irritating lies in the reason: every message has to bring something new, not merely remind.
- The last message is the most effective — the one that explicitly closes the matter.
- What gets automated is the reminder and the draft, not the sending. The sender stays human.
Following up has a bad reputation because it is mostly done badly: the same question in different words, at ever shorter intervals. That genuinely is irritating — but it is the execution, not the following up.
What separates helpful from irritating
| Irritating | Helpful |
|---|---|
| "Just checking in …" | a new reason: an example, an answer, a pointer |
| Even intervals, closely spaced | widening intervals |
| No end in sight | an announced final attempt |
| Pressure ("only until Friday") | relief ("no problem if the timing is wrong") |
| Sent automatically, impersonal | approved by a human, referring to the conversation |
Worth knowing
The most effective message in the whole sequence is the last one — the one where you explicitly close the matter: "I assume the timing is not right, so I will not follow up again. If that changes, do get in touch."
The reason is easy to follow: that message takes the pressure off and makes replying easy, because it asks for nothing. In practice it regularly produces more responses than the three before it combined — and the reply is often that it is still relevant after all.
The four-step sequence
Day 3 – The follow-on
Reason: something left open in the conversation — a figure, an example, an answer to a question.
Length: under 80 words. Question at the end: one, concretely answerable.
Day 8 – The parallel
Reason: a case from your own work that matches the contact's situation. No offer.
Length: under 120 words. Question at the end: optional.
Day 18 – The relief
Reason: explicitly acknowledge that the timing may be wrong, and ask when a better moment would be.
Length: under 60 words.
Day 40 – The close
Reason: close the matter, without reproach and without a final question. Leave the door open.
Length: under 50 words. Afterwards: back into general communication, no further following up.
What gets automated — and what does not
| Step | Automatic | Human |
|---|---|---|
| Setting deadlines and reminding | yes | – |
| Preparing a draft referring to the conversation | yes | – |
| Choosing the reason | suggestion | decision |
| Sending | – | yes |
| Stopping on a reply | yes | – |
| Stopping on a decline | yes | – |
The reason fully automatic sending does not work here is not politeness but the reason itself: the value of each message lies in bringing something new — and what is new and fitting comes out of the conversation.
A process that takes over the reminder and the draft and leaves the choice to the human costs about two minutes per message. That is the difference between a sequence still being read after four messages and one that lands in spam after the second.
Stop rules
Four conditions end the sequence immediately:
- The contact replies — whatever they say.
- The contact declines or names a different time. Then it becomes a reminder rather than a follow-up.
- The messages are not arriving — after two delivery failures, stop and check the address.
- The contact has unsubscribed. That applies to personal-looking messages too, if they come from a system.
Draft a follow-up message for me. I will give you the context – do not invent anything on top of it. Situation: - What was discussed in the first conversation: [note] - What was left open: [point] - How long ago that was: [days] - Which attempt is this: [1 of 4 / 2 / 3 / 4] - What I have that is new since then: [example, figure, an answer to the open question – or "nothing"] Rules: 1. The message has to have a reason of its own, not merely remind. If I entered "nothing" above, tell me so and suggest what reason I could go and get. 2. No pressure, no deadline, no "only until". 3. At most one question, concretely answerable. 4. Length by attempt: 1st under 80 words, 2nd under 120, 3rd under 60, 4th under 50. 5. On the fourth attempt: close the matter explicitly, without reproach, leave the door open, ask no question. Give me two versions to choose from. Use no detail that is not above.
In closing
Four attempts at widening intervals are not intrusive — provided each brings something new and the last one explicitly closes the matter.
What belongs automated is the reminder, the draft, and above all the stop on a reply. The sending stays with the human, because the value of the message lies in its reason — and that comes from the conversation, not from a rule.
Common questions
How often may you follow up?
Four attempts at widening intervals are defensible — roughly day 3, day 8, day 18 and day 40. What matters is not the number but that every message has a reason of its own and that the last one explicitly closes the matter.
What separates helpful from irritating follow-ups?
The reason. A message bringing something new — an example, a figure, an answer to an open question — is helpful. One that merely reminds is irritating. Add widening rather than even intervals, no artificial time pressure, and an announced ending.
Which message works best?
The last one — where you close the matter and ask for nothing. It takes the pressure off and makes replying easy; in practice it often produces more responses than the three before it combined, and the reply is frequently that it is still relevant after all.
Can following up be fully automated?
Not usefully. What can be automated is deadlines, reminders, drafts and above all stopping on a reply. Choosing the reason and sending stay with the human — the value of each message lies precisely in picking up something fitting from the conversation.
When does the sequence have to stop?
On any reply from the contact, whatever it says; on a decline or the naming of a later date — then a reminder replaces the follow-up; after two delivery failures; and on any unsubscribe, even where the message looks personal.
Marketing that sets itself up
The Studio Engine beta is live. Claim your spot and help shape it from the start.
Join the beta →