How many follow-up emails are too many?

My longest sequences run to 8 emails across 3 weeks, and that is the upper limit. More important than the count is the gap (2-3 days) and a hard limit: at most one email per day per recipient, however many sequences they are in. The last email is always a breakup: you say clearly that you will not write again, and you honour it.

When does an automated sequence have to stop?

On every sign that it has done its job or is not welcome: a reply, a booked meeting, a purchase or an unsubscribe. Each of those exits has to be automatic, because stopping by hand is always late. The worst case is a buyer who gets a payment confirmation and a 'writing one last time' email on the same day.

Can I send cold emails to addresses I collected from the web?

No. Unsolicited commercial messages require consent under the EU ePrivacy rules and their Croatian implementation, and there is no clean B2B exemption. My approach is phone-first: the email goes out only after a call and the person's agreement. It is also more effective, because an email that arrives after a conversation gets read, while an email out of nowhere gets deleted.

What counts as good personalisation in a follow-up email?

Real data about the recipient, not their name in the greeting. 'Dear Ivan' is not personalisation, every spam message knows that. Personalisation is 'your webshop scored 61/100 on the check, your biggest competitor 84': a number from a real analysis that concerns the recipient. If you have nothing concrete about them, you are not ready for a sequence.

Which tool should I use for email sequences?

The tool matters less than the rules: automatic exits, a daily limit, a permanent unsubscribe. Off-the-shelf tools such as Mailchimp or Brevo can do it with a disciplined setup. I built sequences into my own CRM so they also shut off on events from other systems, such as a payment or a booked slot. For most businesses that is a project of a couple of weeks.

All postsBlog

Follow-up that shuts itself off: email sequences without spam

How many hours a week manual follow-up costs you, and how sequences take it over: they stop on a reply, a meeting or a purchase, one email a day at most.

Case study
Email
Automation

Jurica StublićFounder · ai.webica.hr8 October 2026·7 min read

Most deals are not lost on the first email. They are lost on the follow-up that never happened.

Someone asked for a quote, got it, and then silence. Not because the quote does not interest them, but because they were busy, the email sank, and nobody on your side got back in touch. Manual follow-up dies in the first hard week. Automated follow-up survives every week.

But automated follow-up has a dark side, and we have all seen it from the buyer's side: emails that keep arriving after you have replied, "just following up once more" number seven, offers for something you already bought. That is not automation, it is spam on a schedule.

In this article I go through how I built sequences in my own CRM that do the first part and not the second. All the rules transfer, whatever tool you use.

The anatomy of a sequence

A sequence is a pre-written series of emails with gaps between them. My longest, for webshops after a free compliance check, is 8 emails across 3 weeks: day 0, then every 3 days.

Three rules hold the quality of the series:

One email, one topic, one call to action. An email offering three things offers nothing.
Every email stands on its own. A recipient who sees only the fourth email has to get something useful, not "as I wrote last time".
The last email is a breakup. You say clearly that you will not write again. And you honour it. Paradoxically, the breakup email often pulls the most replies, because a deadline forces a decision.

Personalisation: real data or nothing

"Dear Ivan" is not personalisation. Every spam message in the world knows that trick.

Personalisation that works is a concrete number that concerns the recipient. My emails for webshops pull data from their actual check: how many points the shop scored, which legal items it is missing, how its biggest competitor stands. The sentence "your shop scored 61/100 on the check, your competitor 84" does work no copywriting can.

From that follows a readiness test: if you have no concrete data about the recipient, you are not ready for a sequence. Analysis or a conversation first, then emails.

How to arrange topics across a sequence

Eight emails does not mean eight rounds of "have you had a chance to think about it". Every email carries its own value, and the series has a dramatic arc:

The beginning (day 0-3): what you promised. The analysis, the numbers, the answer to the question that put the recipient in the sequence in the first place.
The middle (day 6-15): one concrete insight per email. The most common mistake in their situation, a short example from practice, the answer to the most common objection, the arithmetic of the cost of waiting.
The end (day 18-21): a summary of the offer, and the breakup.

The test for every email in the series: would a recipient who read it in isolation, without any of the others, get something useful? If the answer is no, that email is a reminder, and reminders get deleted.

To give a sense of it, here is the skeleton of a breakup email, the shortest and most important in the series:

Subject: Closing your file

I have been in touch a few times about [concrete topic] and I do not want to be the person sending a tenth reminder. This is the last email: if it becomes relevant, my number and the analysis are still good in three months. Thanks for your time.

Three sentences, the exit left open, dignity preserved on both sides. And that is exactly why it works.

The most important part: when the sequence shuts off

The quality of automated follow-up is determined not by what it sends, but by when it stops.

My sequences shut off automatically on each of these events:

A reply to an email. A person got in touch, the machine steps aside, I take over the conversation.
A booked meeting. The calendar tells the system and the sequence stops. Reminding someone who has already booked a slot is pointless.
A purchase. Payment kills the sales emails on the spot.
An unsubscribe. No questions, no "are you sure".

The key word is automatically. Stopping by hand is always late, because it relies on someone seeing a reply and remembering to switch the machine off. The worst possible scenario — a buyer who gets a payment confirmation and a "writing one last time" email on the same day — happens in exactly those systems where sales and email are separate worlds.

So in mine, sequences live in the same system as payments, the calendar and the inbox: an event anywhere in the funnel shuts off emails everywhere.

Guardrails against spam

Alongside the exits, three hard limits:

At most one email per day per recipient. This holds even when someone happens to be in two sequences: the system counts what has been sent today and simply does not send the excess.
An unsubscribe lasts forever. Anyone who unsubscribed does not come back onto the list through a new contact import, a new campaign, or anything else. That list is never deleted.
Sending has a rhythm. Emails go out spaced apart, not in bursts. Bursts are the fastest route to the spam folder, because bursts are exactly how mail providers recognise bulk mail.

None of this reduces sales. The only thing it reduces is the number of people you have annoyed, and for the reputation of your domain and your brand that is pure profit.

Technical reliability, briefly

Two things you only notice when they fail. A sequence running for three weeks has to survive everything that happens to the software in those three weeks: upgrades, restarts, outages. So every step is held as a durable record, not as a timer in memory that disappears with a restart.

And second: sending has a lock. When two processes reach for the same email at once — say the automation and a manual click — only one gets through. A duplicate email is a small thing that costs more trust than the whole sequence is worth.

A sequence is not a newsletter

It is worth separating two things that often get confused, because they have different rules.

A sequence is a personal series tied to one recipient and their moment: it started because they did something, and it shuts off when they do something. A newsletter or broadcast is one message to everyone on the list: news about a new service, an article, a change in the law that concerns everyone.

I have both in the same system, and they share protections: the same unsubscribe list applies to everything, the same sending rhythm without bursts, the same limits. But their tone differs. A sequence may be personal and specific because the recipient is in a conversation with you. A broadcast pretending to be a personal email ("sending this just to you") is a trick everyone recognises and that spends trust.

When you are unsure which is which, the question is: does the recipient's status shut the message off? If yes, it is a sequence. If no, it is a broadcast.

Consent: email only after a call

To close, a limit that is not technical but legal. An unsolicited commercial message requires consent under the Zakon o elektroničkim komunikacijama (the Croatian Electronic Communications Act), which transposes the EU ePrivacy rules that apply across the Union, and there is no comfortable B2B exemption for it. A cold blast to addresses scraped from the web is unlawful, and useless besides.

So my order is always the same: a call first, then, with the person's agreement, an email with a concrete analysis, then the sequence. An email that arrives after a conversation gets read. An email out of nowhere gets deleted and reported.

A checklist for your follow-up

If you put only this in place, you are ahead of 90% of the market:

Defined automatic exits: a reply, a meeting, a purchase, an unsubscribe
Limits: 1 email per day per recipient, 6-8 emails in total at most
A breakup at the end, and silence after it
At least one real piece of data about the recipient in every email
An unsubscribe that lasts permanently

The tool matters less than the rules. Off-the-shelf tools can do this with discipline, and building it into your own system, as I did in my CRM, buys you the most valuable part: shutting off on events from across the whole funnel, from a payment to a slot in the calendar.

Summary

Follow-up has to be automatic to survive, and clever enough not to become spam. The formula from my system: up to 8 emails across 3 weeks with a breakup at the end, personalisation with real data from an analysis, automatic shut-off on a reply, a meeting, a purchase and an unsubscribe, a hard limit of one email a day, and an unsubscribe that lasts forever. Plus an order that respects consent law: conversation first, then email.

If you do not know how many hours a week really go on follow-up and other repetitive work, that can be measured. The process analysis is two meetings walking through how you work today, ending with a written report that ranks jobs by what pays off first to automate. For quick first ideas without a conversation there is also the free AI analysis.

About the author

Jurica Stublić

Founder of webica.hr and ai.webica.hr

4+ years of development and AI automation. I work alone: you talk to me from the first conversation through to launch. I bring experience from 50+ web projects at webica.hr to AI automation for companies of 5 to 50 people.

See the process analysis More about me