Yes, with one asymmetry. An agent can act on any request to receive less marketing straight away, because reducing contact is always safe. Opting a customer in needs clear, recorded consent. And some messages cannot be switched off by preference at all, such as fraud alerts, required notices and arrears contact, where a person decides how contact happens.
Why is an opt-out easier than an opt-in?
An opt-out removes a permission. If the agent gets it wrong, the customer misses some offers. An opt-in creates a permission, and marketing on a permission the customer never gave is a data protection problem. Under GDPR, marketing usually rests on consent or another lawful basis, and the business must be able to show it.
So the agent applies opt-outs immediately and confirms them. For opt-ins, it records exactly what the customer was asked, what they said and through which channel, so the consent can be evidenced later.
Which messages can a preference not switch off?
Customers often say "stop contacting me" when they mean marketing. Some contact is not optional for the business: security alerts, notices it must send by law or contract, statements, and messages about missed payments. The agent explains which messages will continue and why, rather than promising silence.
Arrears contact is the hard case. A customer in financial difficulty may ask for all calls to stop. The business may be able to change the channel, time or frequency, and may need to, but the decision belongs to the collections team, who weigh the customer's circumstances.
What may the agent do, and what stays with a person?
| Request | The agent | A person |
|---|---|---|
| Stop marketing emails, texts or calls | Applies the opt-out and confirms | None |
| Start receiving offers | Records consent wording and applies it | Samples consent records |
| Switch statements from paper to email | Applies it, where the product allows | Exceptions for customers who need paper |
| Only contact me by email, never by phone | Applies it for service messages where possible | Decides for arrears and fraud contact |
| Stop all calls about my overdue balance | Records the request and the reason | Agrees a contact plan with the customer |
| Contact my carer instead of me | Records the request, takes no action | Handles authority and vulnerability |
What does this look like in practice?
Take a mobile network. A customer writes: "Stop calling me. I've had four calls this week." The agent checks the account: two calls were marketing, two were from collections about a missed payment.
The agent opts the customer out of marketing calls at once and confirms it. On the collections calls, it does not promise they will stop. It explains that the calls relate to the missed payment, asks whether email or a set time would work better, and records the answer and the customer's mention that they lost their job last month. That note goes to the collections team, who treat it as a possible hardship case and agree a new contact plan. The agent has stopped what it may stop and passed on what it may not.
Which controls make it safe?
- Opt-outs applied everywhere they matter: the marketing platform, the dialler and any outsourced callers. A preference that stops at one system is not honoured. Joining preferences to calling systems is covered in how to connect an AI agent to your contact centre telephony.
- Consent captured with its wording, channel and time.
- A fixed list of messages that preferences cannot switch off, read from the policy library.
- Arrears, fraud and vulnerability requests routed to the right team.
- Confirmation to the customer of every change.
These requests are usually quick to resolve in one contact, which helps your first-contact resolution. Tracking how often customers ask to stop contact also shows where your outbound messages cause friction; see how to report on contact drivers with an AI agent.
Questions people ask
Does changing contact preferences need the same checks as changing an address?
Less for opt-outs, since they reduce risk. Changing the email or phone number itself is different and is covered in whether an AI agent can change a customer's address.
Can the agent move a customer to a cheaper paperless plan at the same time?
It can explain the option. Moving the plan is a separate capability; see whether an AI agent can change a customer's plan or tariff.
What if a customer asks to stop all contact after a bereavement?
The agent stops marketing at once and hands over to a person trained for bereavement, who handles the account from there. The wider approach is in the guide to AI customer support agents.