Which processes to automate first: a checklist for Malaysian SMEs

The most common reason an automation project disappoints is not bad engineering. It is that the wrong process was picked to start with. Usually the most visible one rather than the most suitable one.

Score each candidate on four things

Take your three or four most annoying processes and rate each one from 1 to 5.

• Frequency. How often does this happen? Daily beats monthly by a wide margin.
• Rule clarity. Could you write the decision down as rules? If it depends on judgement or relationships, score it low.
• Data availability. Does the answer already exist in a system, or does someone have to go and find out?
• Cost of getting it wrong. Low stakes is good for a first project. Do not start with something where a mistake is expensive.

Add the scores. The highest total is usually your starting point, even if it is not the process that annoys you most.

Good first candidates

• Status enquiries. Order status, delivery status, stock levels. The answer exists, the delay is entirely in someone looking it up.
• Leave and claims. Clear rules, high frequency, low ambiguity.
• Recurring reports. Same format, same sources, produced on a schedule by a person who resents it.
• Data re-entry between two systems. The clearest waste there is.

Poor first candidates

• Anything involving negotiation or relationship judgement.
• Processes nobody can explain the same way twice. Fix the process first.
• One-off exceptions. If it happens four times a year, automating it will never pay back.
• Anything where a mistake damages a customer relationship or creates legal exposure.

Measure before you start

Spend a week counting. How many times does this happen, how long does each one take, and who does it. Without a baseline you will not be able to tell whether the project worked, and neither will anyone you have to justify it to.

The number is usually worse than people expect, which is helpful when you need a decision.

Start narrow

One process, one department, one clear measure of success. A working system covering a small area beats a broad rollout that nobody trusts. You can widen it once people believe it.

See what this looks like across five departments.