WhatsApp and customer care

Automating WhatsApp Without Losing Control: Consent, the 24-Hour Window, Templates and Human Escalation

WhatsApp can be automated, but the platform sets clear rules. We explain how consent, the 24-hour window, templates, privacy and human handoff fit into a professional customer care system.

9 September 2026 · Updated on 21 September 2026 · 7 min read · My NovaGen AI team

Automating WhatsApp is more than connecting a chatbot and letting it reply to messages.

WhatsApp Business has specific rules on how a company can contact someone, when it can reply freely, and when it must use an approved template.

If automation is involved, Meta's policy expects a clear path to a human agent.

The practical takeaway for any business: automation must work alongside consent, context, human control and proper data management.


WhatsApp is the visible channel, but the business must control consent, context and the type of messages it sends.
WhatsApp is the visible channel, but the business must control consent, context and the type of messages it sends.

Under the WhatsApp Business policy, contacting users requires their prior consent. Check the current wording before relying on it.

That consent must make clear that the person agrees to receive messages from the company, and it must name the company.

So WhatsApp should not be treated as a mailing list just because you have someone's number.

A serious system must know:

  • where the contact came from;
  • what consent exists;
  • what type of messages the person expected to receive;
  • whether they have opted out;
  • when the commercial context stopped being valid.

2. The 24-hour customer service window changes what the business can send

The customer service window lasts 24 hours from the customer's last message. Inside it, the business can reply freely. Outside it, the business can only message using approved templates.

Inside that window, replies do not need a message template.

Outside it, business-initiated messages need approved templates and may carry a cost.

This shapes how automations are designed.

For example, a follow-up flow cannot assume it can message freely several days later with any text.

Automation needs to know whether the conversation is still inside the window and act accordingly.


3. A template should not become automated spam

Templates exist so businesses can start conversations within the platform's rules.

But an approved template is not automatically a good idea to send without context.

The system should check:

  • whether consent exists;
  • whether the message is relevant;
  • whether the user has already replied;
  • whether they have asked not to receive more messages;
  • whether the customer's status has changed;
  • whether the template actually fits the goal.

Automating a bad decision only makes it happen faster.


When automation should not carry on alone, the handoff to a person must be quick, clear and come with enough context.
When automation should not carry on alone, the handoff to a person must be quick, clear and come with enough context.

4. Automation is allowed, but escalation must exist

Inside the window, the policy allows automated replies, provided the required safeguards are in place.

It also requires timely, clear and direct escalation routes.

Meta lists, among others:

  • transfer to a human agent;
  • telephone;
  • email;
  • online help;
  • a support form.

This should show up in the actual user experience.

A button buried somewhere on a web page is not enough.

If the user asks for a person, or the AI detects that the request is outside its scope, the handoff must be simple.


5. The agent must know when to stop replying

In several cases, pushing on with automation makes the experience worse.

Complaint or dispute

An upset customer usually needs context and human judgement.

Sensitive data

Not all information should be handled in the same flow.

Commercial exception

A special discount, a refund or a term outside the standard catalogue may need approval.

Repeated failure

If the system no longer understands the conversation, persisting only creates frustration.

Explicit request

If someone asks for a human, they should not have to “win” against the chatbot.


Privacy, access and data protection should be built into the WhatsApp flow from the design stage, not bolted on at the end.
Privacy, access and data protection should be built into the WhatsApp flow from the design stage, not bolted on at the end.

6. Privacy: collect only what is needed

The Spanish Data Protection Agency (AEPD) stresses data minimisation: processing only the data that is adequate, relevant and necessary for the purpose.

In a WhatsApp agent, this principle is easy to apply in practice.

If booking an appointment only needs a name, phone number, service and time slot, there is no reason to ask for more out of habit.

It is also worth checking:

  • how long data is kept;
  • which data flows into the CRM;
  • which data reaches the AI model;
  • who can see the conversation history;
  • what happens when the customer asks to stop receiving messages.

Data protection is designed into the flow, not tacked on at the end as a legal notice.


7. WhatsApp and the CRM must share status

Imagine a lead is marked “Do not contact” in the CRM.

The agent should not send a sequence just because an outdated rule says so.

Or imagine someone has already accepted a proposal.

It makes no sense for them to keep receiving messages written for someone still weighing the decision.

That is why systems need to share relevant customer status data.


8. Automating follow-up does not mean bombarding

Good follow-up can add real value.

But it has to be limited.

For example:

  1. An enquiry comes in.
  2. It gets a reply.
  3. If information is missing, ask for it.
  4. If an action is still pending, a reasonable reminder is scheduled.
  5. If the customer replies, the pending reminder is cancelled.
  6. If they ask not to receive more messages, the sequence is blocked.

The key: every message needs a reason.


9. Traceability protects the customer and the business

Keep a record of:

  • when the message arrived;
  • which system replied;
  • which template was used;
  • which action was run;
  • whether a person intervened;
  • whether an error occurred;
  • whether an opt-out happened.

Not to monitor conversations to collect data, but to understand what happened when something goes wrong.


10. A complete example

A customer asks for information outside opening hours.

The agent answers their questions and realises they want an appointment.

It checks availability and offers two slots.

The customer does not reply.

If the conversation is still inside the window, the flow can carry on; if not, an approved template is needed.

Three days later, if the company wants to send a new message, the system must check whether an approved template applies and whether there is enough consent for that type of communication.

If the customer asks for a person at any point, the conversation is handed over and the automation stops competing with the team.

To the user, this looks simple.

But it requires state, rules and control.


Practical summary of control on WhatsApp

In practice: before sending an automated message, the system should know why it is allowed to send it, what state the conversation is in, and whether any condition means the sequence must stop.

SituationUseful check
The customer has just messagedConversation context and a suitable reply.
Several days have passedWhether an approved template applies and there is a valid reason to get in touch.
The customer has already repliedCancel any pending reminders.
The customer asks not to receive messagesBlock new sequences and keep that status recorded.
They ask to speak with a personHand over and stop automation competing with the team.

Frequently asked questions

Can WhatsApp Business be automated?

Yes, within the platform's rules and with a design that covers consent, templates where needed, privacy and handoff to a person.

What does the 24-hour window mean?

It is the customer service period tied to the user's conversation, during which certain messaging rules apply. Outside it, the business must check what kind of message it can start and whether it needs an approved template.

Should an automated reminder still go out if the customer has already replied?

No. The conversation and CRM status should cancel any action that no longer makes sense.

You can explore this further in our AI WhatsApp agents service or request an initial assessment.

Conclusion

The best WhatsApp automation is not the one that sends the most messages.

It is the one that knows:

  • when it can reply;
  • when it needs a template;
  • what consent exists;
  • what data it may process;
  • when it must stop;
  • when it must pass the conversation to a person.

That turns WhatsApp into a well-run part of the customer care system, rather than yet another channel that is hard to control.


Further reading

If you are still deciding what is worth automating, start with our practical guide to automating WhatsApp in a business. Here we go deeper into the terms of use and how you control outgoing messages.

Reference sources

Share article

Does any of this sound like your situation?

We can review your specific case and tell you, with no commitment, what is worth automating and what is not.

Talk to Nova on WhatsApp