The two gates every outbound message passes
- Consent gate: is this contact reachable on this channel for this purpose?
- Permission gate: is this person or automation allowed to send on the company's behalf?
- Both must pass. A permitted sender may not message a contact who opted out.
- Transactional messages about work in progress are treated differently from marketing — the distinction should be explicit in your system, not assumed.
What a correct opt-out looks like
It takes effect immediately, on the channel used, without a person having to action it. It is visible on the contact record so nobody wonders why messages are not sending. It blocks automations as well as people — the most common failure is a human stopping while a sequence keeps going. And it survives a spreadsheet import, which is where suppression lists usually die.
Re-consent must be an explicit act by the customer. Nobody in your office should be able to click a contact back into reachability.
Permissions as an operating control
Not everyone should be able to text every customer. A field technician confirming arrival is different from a salesperson sending pricing, which is different from an automation running a follow-up sequence at 8pm. Deciding these in advance prevents the two situations that damage businesses: nobody sending anything, and everyone sending everything.
AI and automated outreach deserve the tightest gate of all, because volume is what turns a small policy mistake into a large one.