Three automated emails count as mass advertising in Switzerland. A hundred sent by hand do not.
That is not a technicality. It is the test the Swiss Commission for Fairness in Commercial Communication applied in 2021, and it reverses the order most companies work in. The first question is not which tool to buy. It is at which step a different set of rules starts applying to you.
What marketing automation actually is sits in the glossary. This piece is about the decision either side of it: which processes genuinely pay off for a smaller company, in what order they should be built, and how you notice you have gone too far.
Automation is the trigger, not the volume
Electronic mass advertising is lawful in Switzerland only when three conditions are met together: consent was obtained beforehand, the sender is correctly identified, and there is a straightforward, free way to opt out. That is Art. 3 para. 1 lit. o of the Federal Act against Unfair Competition (UCA, SR 241).
Any serious tool handles two of those three on its own. Not the first.
The decisive part is what "mass" means. The Swiss Commission for Fairness settled that in decision no. 172/20 of 3 August 2021: the test is whether sending is automated, not how many people receive it. A single automated send to a handful of addresses is covered – and a message typed by hand to a long list is not.
That is the sentence this article turns on: the moment you automate sending, you move into a different legal frame. Not above a thousand addresses. At the first sequence.
This is not hypothetical. In 2019 a sender in the canton of Lucerne was convicted over 22 advertising emails to an address whose owner had explicitly objected: a suspended fine of 15 daily units at CHF 70, plus a CHF 250 penalty and CHF 660 in costs (Steiger Legal, 2019).
22 emails. Not 22,000.
The existing-customer exception is narrower than it sounds
The act does provide a route: if you obtained contact details in connection with a sale, you may advertise your own similar goods or services. The conditions are cumulative, though – all of them or none.
- The data came from a sale, not from a contact form, a business card or a download.
- What you advertise is your own and similar. "Similar" is read narrowly and means an obvious connection to what was bought, not your whole range.
- The opt-out was pointed out at the moment the data was collected, not in the first email.
If you cannot evidence all three, you are working on consent. And a consent that only lives in someone's memory is not one: who, when and what for belongs in the CRM, with the date attached.
The four sequences that pay off first
Automation earns its keep where a task happens often, follows clear rules and needs no case-by-case judgement. All three together, not two of them.
Apply that test and four candidates survive for a smaller company – in this order.
1. The reply to an enquiry
The best ratio of effort to effect. Someone submits the form, gets an immediate confirmation with a realistic response time, and the enquiry lands with the right person at the same moment.
Legally this is the easy case: answering an enquiry is not advertising, it is delivering what the person handed over their address for.
What this sequence really improves is not speed but reliability. An enquiry arriving on a Friday evening otherwise looks like silence until Monday.
2. Appointments: confirmation, reminder, follow-up
Second, because it prevents a measurable loss instead of promising an assumed gain. Confirmation immediately, reminder the day before, a short summary with the next step afterwards.
No-shows are the only quantity on this list you can count before and after. That makes this the one sequence that proves itself.
3. The handover to sales
Here you are not automating outbound communication but an internal notification – which removes the entire legal question above.
When someone views the same service page repeatedly or opens the pricing page, the responsible person gets a message. They then decide whether and how to reach out.
Automation does not replace sales, it aims their attention. That is exactly the line between a useful system and an embarrassing one: personal contact stays personal.
4. Onboarding after the order
The most underrated case, because it sits after the sale and therefore rarely appears in a marketing budget.
If the same documents, access details and instructions go out after every order, that is a sequence. It saves time, and it stops something being forgotten with the tenth client.
What usually does not pay off
Three things that look good in every product demo and rarely hold up in a small company.
Branching sequences with many conditions. A flow with fifteen forks has to be maintained, and it only gets maintained while someone still remembers how it was meant to work. Six months on it sends things nobody checks.
Lead scoring at small numbers. Points for clicks and page views only form a picture once enough cases accumulate. At fifteen enquiries a month you know your prospects by name anyway – scoring is then effort without insight. How to generate enquiries predictably in the first place sits in the marketing funnel.
The sequence meant to compensate for missing content. A run of five weak emails just delivers weakness more reliably. Automation amplifies what exists; it does not create anything.
Order matters more than feature set
The most common mistake is not the wrong tool. It is too large a first step.
A workable route looks like this:
- Name a bottleneck that actually hurts. Not "we want to automate", but "enquiries from the weekend sit untouched for two days".
- Build one single sequence that solves exactly that bottleneck.
- Run it for four weeks and measure against the state before.
- Only then the next one.
Step three is the one that gets skipped. An automation without a baseline cannot be switched off later, because nobody can say whether it works. Which metric that is, you decide beforehand – not once the numbers are in.
For the internal half you often need no marketing software at all: if you already run Microsoft 365, notifications, handovers and filing are covered by Power Automate without a second subscription (what that looks like for a smaller company).
How you notice there is too much of it
There is no warning light. But there are three signs, and all three are readable.
The unsubscribe rate climbs while the list stays the same size. That is the most honest feedback there is: people leave before you asked.
The replies stop. If nobody ever writes back to an automated message, nobody is reading it either.
You no longer know what is going out. If nobody in the company can say without checking which email a prospect received this week, the sequence has stopped being relief and become a blind spot.
The test is simple: every automated message needs a purpose the recipient can recognise. An appointment confirmation has one. A reminder has one. The third email inside a week rarely does.
What comes before automation
Automation is the last step, not the first. It assumes three things no software brings with it.
Enough enquiries that differentiation is worth anything. At five enquiries a month a personally written email beats any sequence. What generates enquiries sits in the twelve levers of website lead generation.
Evidenced drop-off points rather than assumed ones. Which steps deserve automating only becomes visible once you know where people leave – and that is a question of data, not imagination. How to find those points sits in customer journey mapping.
A page that carries the step. The best sequence is worthless if the landing page behind it does not lead to an enquiry, or nobody finds the page through search.
The calculation before you build anything
An automation pays off when it saves more time than it costs to maintain. That is calculable, with numbers you already have.
Take a task you want to automate. How often does it occur per month? How many minutes does it take by hand? That is the saving.
Against it sit the build – one-off – and the upkeep. Do not price the upkeep at zero: a sequence nobody reads through twice a year eventually sends something wrong.
If the saving does not cover the effort within the first year, the answer is not "automate later" but simplify the process first. A bad workflow does not get better through automation, only faster.
In short: not as much as possible, but the right things
Marketing automation for a smaller company need be neither complicated nor expensive. What matters is which processes you pick – and knowing what changes legally with the first automated send.
The four sequences that almost always hold are the reply to an enquiry, the appointment chain, the internal handover to sales, and onboarding after the order. Everything beyond that is a question of scale.
And the one rule that stays: automate not as much as possible, but the right things. Whatever repeats, follows clear rules, and disappoints nobody when it happens without a person.