Growth creates more work.
More customers. More enquiries. More approvals. More coordination.
At some point, the team starts spending too much time just keeping processes moving.
That is when founders usually ask: What should we automate first?
In this article
- Find the bottleneck
- Decide what should stay human
- Prioritise by impact and risk
- Check automation before adding headcount
1. Start with the bottleneck, not the technology
Do not begin with: “Where can we use AI?”
Begin with: “Where is work getting stuck?”
Look for places where:
- customers are waiting
- teams keep chasing each other
- information is entered twice
- approvals slow things down
- reports take too long
For example, if quotations take two days because pricing comes from multiple systems, the goal is not “automate quotations.”
It is: Remove the delay between enquiry and quote. ⏱️
2. Decide what machines should do — and what people should do
Useful automation rarely removes the whole process. It removes the repetitive part.
Take a sales enquiry. Automation can:
- collect the enquiry
- check CRM history
- identify the company
- suggest a lead score
- prepare a response
A salesperson can then decide whether the opportunity is worth pursuing.
AI is often strongest at preparing the decision, while people handle important judgement. 🧠
3. Prioritise by impact and risk
Do not automate everything at once.
Compare opportunities using:
- Time saved
- Frequency
- Business impact
- Risk if it goes wrong
| Process | Automation priority |
|---|---|
| Repetitive reports | High |
| Qualifying enquiries | High |
| Drafting standard quotations | High |
| Approving large refunds | Medium |
| Hiring decisions | Low |
The best first automation usually happens often, saves visible time and is easy to verify. ✅
4. Check the process before adding another person
Growing workload often leads directly to hiring. Sometimes that is right.
But if a new employee would mainly spend time:
- gathering information
- preparing routine documents
- updating systems
- checking statuses
it is worth asking whether part of that workload can disappear first.
The goal is not to replace people. It is to avoid hiring people just to manage inefficient processes.
A simple place to start
Pick one process that regularly causes delays.
Ask: What part of this process actually requires a person?
Keep that part human. Automate the rest where it makes sense. 💡