No regulator requires one opt-out to stop every channel. Do it anyway, and not for the reason you think.

The law probably does not make you. You should probably do it anyway.

Those are two different answers to two different questions. Most of the advice online quietly rolls them into one, which is how people end up believing the law says something it does not.

Here is what the regulators actually say, what the big platforms actually do, and why we still tell people to take the stricter path if their business runs on messaging.

01 / The uncomfortable partThe uncomfortable finding first

Nobody in the industry agrees that saying no on one channel should stop every channel. What they DO agree on is three things.

A "no" applies to the channel it was said on. You must always offer a "stop everything" option next to it. And if someone asks you to stop entirely, that is final, with no exceptions.

One piece of vocabulary before we go on, because you will meet it everywhere. The trade word for not sending to someone is SUPPRESSION. It means keeping them on file and never messaging them, which is not the same as deleting them, and that difference turns out to matter a lot further down.

Two regulators write the one-channel rule in almost the same words. Every major email and customer-database platform is built that way underneath. And the one regulator that tried to REQUIRE the wider version, in the US, has put its own rule back twice, on the grounds of cost and effort, currently to 2027-01-31.

If your gut says "a no anywhere should mean no everywhere", your gut is making a fair business decision. It is not the legal minimum, and it is not what the industry does.

Know which of the two you are arguing. One day somebody will ask you to justify it.

02 / The mapThe map

Read the Singapore line twice, because it catches people out. If you reach someone through their phone number, the do-not-call rules apply. If you reach the same person through their social media profile, they do not.

That is the do-not-call rules only. Singapore has separate spam law, and whether it covers messaging apps is genuinely unsettled. We found well-regarded lawyers flatly contradicting each other on it. So do not read that row as covering everything Singapore requires.

We owe you something about the Australia row. Every other row in that table, we read at the regulator or in the law itself. Not that one.

The Australian government's own site and the usual free copy of the case law are both blocked from where we sit. So that row rests on summaries written by law firms, not on the Act.

We think it is right. We have not checked it ourselves. And a table that mixes the two without saying so is exactly the thing this page is arguing against.

03 / Seven rulesThe seven rules that actually govern this

1. A "no" covers the channel it was said on, and every kind of message on that channel. Not only the campaign that set them off.
2. If someone says "stop contacting me entirely", that is final. It covers every channel and you cannot refuse it. They do not have to use any special wording. They can say it out loud, in a private message, to anybody who works for you.
3. Always offer one switch that turns everything off, next to any list of finer choices. A preferences page with no "none of it" option breaks US law.
4. Say what it covers, right where you offer it. The word "Unsubscribe" on its own tells nobody what they just switched off.
5. If you are not sure what they meant, stop more rather than less. You may send one message, not a sales one, asking what they want stopped. If no answer comes back, stop everything that needs their permission.
6. Stopping marketing must never stop service messages. Build the wall between them first, before you switch any of this on.

An appointment reminder is not marketing. A customer who stops getting theirs is a support problem you made for yourself.

1. Suppress, never delete.

04 / Rule 7Rule 7 is the one people get wrong

The UK regulator publishes the example, and it is perfect. A company deleted the phone number of somebody who had opted out, because deleting felt like the respectful thing to do. Later the company bought a marketing list. The same number was on it. They rang. They broke the law.

Think of a guest list. Crossing a name off is not the same as throwing away the note explaining why they are not coming. Print the list fresh next year and they are right back on it.

Keeping their name on a do-not-contact list is not "using their data for marketing". It is how you obey the rule. Deleting a "no" is the classic own goal, and it never fails on the first contact. It fails on the second, which is why nobody catches it in testing.

05 / Our recommendationOur recommendation, and it is a business argument not a legal one

If your main channel is WhatsApp or SMS, stop everything when someone says no. Not because the law makes you. Because the two mistakes do not cost the same, and they are not even close.

• Stop too much and you lose what that one person might have spent. And you can undo it. They can sign back up.
• Stop too little and you get complaints. WhatsApp scores those complaints against the whole business number, over a rolling period, and then slows down your messages to everyone on it. Not just to the person who complained. It is one neighbour's noise complaint turning the volume down for the entire building.

And the reason people pick when a message reaches them on a channel they never signed up for is "didn't sign up", which happens to be the most damaging reason on the list.

So being a bit too strict costs you one relationship. Being a bit too loose costs you the whole list.

That is not a close call. It is also a much better argument than "it is the right thing to do", because it survives a conversation with someone whose job is counting the money.

One thing we will not do is give you a figure for what a preferences page earns back. We chased every one we could find, and each one led to a supplier's blog post citing nothing.

06 / The identifier failureThe failure that has nothing to do with policy

A do-not-contact list only works if you can match the person against it. On WhatsApp, the thing you match them by can now be missing altogether.

A change reached this market in 2026-07. Since then, someone messaging you on WhatsApp can arrive carrying an internal code and no phone number at all.

We counted. Across 2,500 recent conversations on one account we found 29 contacts like this, starting on one particular date and climbing since. That is 1.2% of what we looked at.

We are deliberately not calling it a percentage of NEW contacts, because new contacts is not what we counted.

Nothing is broken here and nothing was deleted. The platform saved exactly what it was given. But three things follow:

• A do-not-contact list kept OUTSIDE your main system, matched on phone number, cannot find these people at all. That means spreadsheets, import files, and any other supplier's list.
• Importing is where it actually bites. A later import carrying the same person WITH a phone number will not recognise the record that has none. Now there are two of them, and the new one carries none of the "do not contact". That is rule 7 failing again, through a different door.
• Any export or report that assumes everyone has a phone number will quietly drop these people or file them in the wrong place.

07 / Our own correctionAnd a correction we owe against our own reasoning

When we first wrote this up, we said that a system matching on phone numbers "cannot match these contacts, so someone who opted out can be messaged again".

We worked that out by thinking about how the system must be built. We never checked. When we did check, it was wrong.

We went and read a real record: a person who had opted out and had no phone number. The opt-out had landed correctly on every channel. The platform files a "no" against the PERSON, not against their phone number. A missing phone makes no difference to it at all.

The real risk turned out to be narrower, and it sits entirely at the edges where systems meet: outside lists, imports, exports. That is a smaller finding and a sharper one. We would never have got to it by thinking harder.

That lesson is worth more than the finding itself. A story that explains what you are seeing is not proof that the story is true. Go and read the record.