Automation will not repair an unclear process. It will repeat it faster. Before choosing a CRM, email platform, WhatsApp provider or integration tool, define the operating rule the software is meant to carry out.
This is a planning checklist, not legal advice. Consent, direct marketing and sector rules need to be confirmed for your organisation and market before a workflow goes live.
The readiness test
| Question | Ready sounds like | Not ready sounds like | | --- | --- | --- | | Trigger | A named event starts the workflow | “When someone seems interested” | | Permission | The permitted channel and purpose are recorded | “They gave us their number once” | | Owner | One role owns the next human decision | “The system handles it” | | Message | The content has one clear job | A generic sequence for every lead | | Exception | A person takes over when the normal path breaks | Failed steps sit in a hidden queue | | Signal | Success can be observed without a vanity metric | “More automation” |
If two or more rows are unclear, work on the process before buying another tool.
1. Define a trigger you can observe
A useful trigger is a specific event: a form is completed, a booking is confirmed, an invoice is paid or a defined period passes without a response. Avoid triggers based on subjective labels that no one applies consistently.
Write the trigger as a sentence: “When this happens, for this type of person, start this workflow.” If the sentence needs several exceptions, split it into smaller workflows.
2. Record permission before the message
Email and WhatsApp flows are not only copy and timing decisions. They use personal data and may be direct marketing. Malaysia's PDPA gives data subjects a right to require direct marketing processing to stop. Singapore's DNC provisions set requirements for certain marketing messages sent to Singapore telephone numbers, with specific exceptions and consent conditions.
Your workflow design should therefore record the channel, purpose, source and date of the permission your organisation relies on, plus how opt-out requests reach the system. Ask counsel or your data-protection adviser to confirm the rule that applies to the exact message and audience.
3. Keep a human owner
Every automated workflow needs a person who owns the outcome. Automation can route an enquiry, send a confirmation or surface an exception. It cannot decide whether a sensitive, high-value or unusual case deserves a different response unless the business has defined that decision.
Name the owner, the queue they watch and the time window for action. Then decide what the customer sees when that window cannot be met.
4. Design the exception before the ideal path
The useful questions are often about failure:
- What if the record is incomplete?
- What if two systems disagree?
- What if the customer replies with an unexpected question?
- What if the assigned person is unavailable?
- What if consent is withdrawn halfway through a sequence?
An exception without an owner becomes hidden operational debt. Put it in a visible queue, give it a response rule and record how it was resolved.
5. Measure the customer and operating signal
Do not measure an automation by the number of messages it sends. Pair one customer signal with one operating signal. For example: completed bookings and time to confirmation; qualified replies and manual follow-up steps; reactivated customers and opt-out rate.
This keeps the workflow accountable to the experience it creates as well as the labour it removes.
A practical first automation
Start with a frequent, stable process that has a clear owner and a reversible mistake. Enquiry routing, booking confirmation and considered post-purchase follow-up are often easier to govern than a large multi-channel nurture programme.
If you want help turning one process into a buildable workflow, start a conversation.