Done well, automation makes follow-ups more personal, not less
There is a legitimate fear behind this question. Everyone has received one of those email sequences pretending to be handwritten and fooling nobody, and nobody wants to be that company.
But it is worth looking at the real cause of the problem. What makes a follow-up feel impersonal is not automation. It is forgetting.
When someone requests a quote and hears back three weeks later with "just checking whether you are still interested", that is not cold because it is automated. It is cold because the person writing no longer remembers the conversation.
A well-built system does the opposite: it remembers everything, arrives at the right moment, and hands you back the context so you can write two lines that apply to that person and nobody else.
The line that divides what gets automated
This is the rule that settles almost every question in this article.
Automate the trigger, the timing and the record. Write the message that asks for a decision.
Invisible work is where automation shines, and it is exactly what nobody wants to do by hand: knowing who is waiting on a reply, working out when it is time to speak again, preparing the context, and logging what happened.
Visible work, particularly when it asks for a decision or a signature, is better written by a person. Not because a machine could not produce correct text, but because that is where trust is decided.
What actually gives an automated follow-up away
If you want to keep the personal touch, it helps to know what destroys it. None of these signs is caused by automation itself, and all of them are avoidable.
Generic opening. "I hope this finds you well" announces that the writer has nothing specific to say.
Merge fields that break. The classic "Hi [name]" that nobody proofread.
Mechanical cadence. Three messages exactly seven days apart, always at the same time, reads as what it is.
Not stopping when the person replies. The worst error of the lot, because it proves nobody is paying attention.
Timing that makes no sense. A message sent at three in the morning on a Sunday is a confession.
No reference to the previous conversation. If the message could have gone to anyone, it went to nobody.
The four layers, from safest to most delicate
Not everything carries the same risk. This scale helps decide where to start.
| Layer | Examples | Automate? |
|---|---|---|
| Transactional | Receipt confirmation, invoice, booking, meeting reminder | Yes, entirely. Nobody wants these handwritten |
| Operational | Order status, sending documents, chasing missing material | Yes, with a real person as sender |
| Relational | Reviving a stalled quote, checking in after delivery | Automate the alert, write the message |
| Decision | Proposing, negotiating, asking for signature, reactivating a former client | Never automate the text |
Most companies do this backwards. They fail to automate the first layer, where there was nothing to lose, and try to automate the last, where there is everything to lose.
The first step, worth more than all the others
Before any sequence, fix the zero moment.
When someone submits a form, they should immediately receive an automatic confirmation stating what happens next and when. It is not a reply, and it should not pretend to be. It is a signal that the message arrived and that someone exists on the other side.
This matters because the window is short. A 2007 study by James Oldroyd of MIT with InsideSales found that the odds of reaching a contact are far higher within the first few minutes than half an hour later. Years afterwards, a Harvard Business Review article by Oldroyd, McElheran and Elkington audited 2,241 companies and found an average response time of around 42 hours.
An automatic confirmation does not replace the human reply. But it holds the person's attention for however long that reply takes, and sharply reduces the chance they move on to the next company.
There is one condition: if you promise a reply within twenty-four hours, deliver it. An automatic confirmation followed by silence is worse than no confirmation at all.
About the statistic everyone quotes
You have probably heard that 80% of sales require five follow-ups and that most salespeople give up after one.
It is worth knowing where that comes from. Practically every article on the subject points to the same source, Marketing Donut, and the original page says only that various studies, carried out at different times, in different places, by different research companies, repeatedly reveal that result. No study is named.
That does not make the idea wrong. Persistence works, and almost everyone gives up too early. But treat the figure as a rule of thumb rather than as science, and be wary of anyone presenting it to you with decimal places.
What actually creates the personal touch
It is not writing by hand. It is specificity.
A message written from scratch but generic lands worse than a message with a fixed structure and one sentence that applies only to that person.
What works is always one of four things:
- Reference what they said. "You mentioned the September campaign" proves someone was listening.
- Reference what has changed since. On their side or on yours.
- Bring something new rather than repeating the request. An example, a reference, a cheaper alternative.
- Make replying easy. A closed question gets answered far more often than a vague invitation to talk.
Notice that none of these requires writing everything afresh. They require having the context in front of you at the right moment, and that is precisely what a system gives you.
How to build this in a small company
You do not need expensive software. You need four things.
One place where contacts are recorded. It can be a simple CRM or a shared sheet. What it cannot be is scattered across personal inboxes.
A status per contact. Waiting on us, waiting on them, deferred until a date, closed. Without statuses, no automation is possible.
Alerts based on status. If a quote has gone five days without a reply, someone should be told. The alert is automatic; what gets written next is not.
Templates with mandatory blanks. A skeleton that saves time, with two or three fields nobody can send without filling in by hand. It is the simplest way to guarantee specificity without losing speed.
A concrete sequence, for a stalled quote
To make this tangible, here is the shape of a follow-up sequence for a quote sent and unanswered. The alerts are automatic, the messages are written.
Day zero. The quote goes out, with one line saying when you will follow up if there is no reply. That alone removes half the sense of pestering, because the next contact stops being a surprise.
Day three. The system flags no reply. The message is short and carries a closed question, along the lines of whether the figure is within what they had in mind. A question answerable in one word gets far more replies than a general request for feedback.
Day ten. Second alert. Here you do not repeat the request, you add something. An example of similar work, a smaller alternative, or a useful piece of information about their sector.
Day twenty-five. Final contact in the active sequence. A message that acknowledges the silence naturally and hands back control, saying you will leave it there and the door stays open. This message has, curiously, one of the highest reply rates, because it removes pressure rather than adding it.
After that. The contact leaves the sequence and moves to spaced check-ins, once or twice a year, with something of value in its own right rather than another request.
Note two things about this design. The intervals grow, which avoids mechanical cadence. And none of the four messages repeats the previous one, which is the difference between persisting and pestering.
And when automation should stay out
It is worth naming the cases where the best system is no system.
When the client is large and the relationship is long, an automated message can read as inattention even when it is purely operational. Here the internal alert is enough and the rest is done by phone.
When there has been a problem, a complaint or a delay, nothing should go out automatically. A sequence running while a client is unhappy turns a small problem into a serious one.
And when the business has very few clients a year, the return on building this is low. A calendar with reminders and some discipline beats a system for managing twelve contacts.
The rules that keep it human
- Stop everything the moment someone replies. No exceptions.
- Cap the sequence. Three to five contacts, then move to spaced check-ins.
- Send from a real person, with the name and address of someone who exists and replies.
- Vary the interval. Two days, then four, then ten.
- Send during working hours, in the recipient's time zone.
- One specific sentence per message, minimum.
- Make leaving easy. Anyone who does not want to be contacted should be able to say so effortlessly, and that is not just good manners. Electronic direct marketing rules require a simple, free way to object.
What to measure
Three numbers are enough to know whether this is working.
Time to first human reply, usually the most revealing and the easiest to improve. The share of contacts that receive any follow-up, where most companies discover they lose half of them by accident. And reply rate by position in the sequence, which tells you where it is worth stopping.
If reply rates collapse from the third contact onwards, the sequence is too long. If most replies arrive on the third, shortening it would be throwing money away.
How Digito helps
We build the invisible part, which is where the return sits.
We connect your website form to wherever contacts are recorded, so nothing gets lost between inboxes. We set up the immediate confirmation and the follow-up alerts based on each contact's real status. We create the templates with mandatory fields that force personalisation. And we leave measurement running, so you can see response time and follow-up rate without calculating them by hand.
What we do not do is write your decision messages for you. Those stay yours, and that is how it should be.
If you feel you lose contacts through forgetfulness rather than lack of interest, that is fixable, and it usually gets fixed quickly.
Sources
- Oldroyd, J., MIT Sloan School of Management with InsideSales, Lead Response Management Study, 2007
- Oldroyd, J., McElheran, K., Elkington, D., "The Short Life of Online Sales Leads", Harvard Business Review, 2011
- Marketing Donut, article on following up sales leads, origin of the five-contacts statistic, presented without identifying the underlying studies
- Portuguese Law 41/2004 and Directive 2022/1 of the national data protection authority, on the right to object in electronic direct marketing