Many small businesses have dozens of repetitive tasks. Automating the wrong one costs more than it saves, and automating a confused process only makes the confusion faster. This guide gives you a simple way to choose the first one.
What makes a good first candidate
- It happens often. A weekly task has less to gain than a daily one.
- It takes real effort. Count minutes, not impressions.
- It goes wrong sometimes. Errors and delays are part of the cost.
- The rules are stable. If the steps change every month, the automation will need constant changes.
- It can be undone. If a mistake can be reversed easily, the risk is low.
- The exceptions are few and known. If you can list them, you can plan for them.
A prioritisation worksheet
List your repetitive tasks and score each from 1 to 3 in every column, where 3 always means "more reason to automate": it happens often, takes real effort, causes errors or delays, follows stable rules, is easy to undo and has few exceptions. Add the scores. The highest totals are your best candidates.
| Task | How often | Effort each time | Errors or delays | Rules stable | Easy to undo | Few exceptions | Total |
|---|---|---|---|---|---|---|---|
| Send booking reminders | 3 | 2 | 3 | 3 | 3 | 3 | 17 |
| Copy enquiries into the job list | 3 | 2 | 2 | 3 | 3 | 2 | 15 |
| Write custom quotes | 2 | 3 | 2 | 1 | 2 | 1 | 11 |
| Reply to complaints | 1 | 3 | 2 | 1 | 1 | 1 | 9 |
These tasks and scores are invented to show how the worksheet works. They are not benchmarks. Your own numbers matter, and a week of honest timing is more useful than a guess.
A process in four steps
- List and time. For one week, note each repetitive task and how long it takes.
- Score. Use the worksheet and choose the top one or two.
- Simplify first. Write the steps down and remove any that do not need to exist. Decide what happens in each exception.
- Automate and measure. Record the time and errors before and after. If you cannot show a difference, the automation was not worth its complexity.
Why a confused process should not be automated yet
If two people do the same task differently, automation has to pick one way, and someone will be unhappy. If nobody can explain why a step exists, it may be unnecessary. Automating it adds software to maintain around a step that should not be there. Start by making the process clear. Often that alone removes much of the effort.
Tasks to leave alone for now
- Tasks that happen rarely.
- Tasks that need personal judgement, such as how to respond to a sensitive complaint.
- Tasks where the rules change often.
- Tasks where a mistake would be costly and hard to reverse, unless a person approves the result.
- Tasks that exist only because of another problem you could fix instead.
Start with what you already have
- Time your tasks for one week.
- Fill in the worksheet.
- Check whether the tools you already use have reminders, templates or built-in automations.
- Write the chosen process as a short numbered list and test it by hand.
When your top candidate is not simple
Ask for help when your top candidate involves several systems, when exceptions are many or when a mistake would affect customers or money. A short review can often show that a simpler process, or a setting you already have, removes the need to build anything. If AI seems attractive for the task, AI or ordinary automation? helps you decide. For the common case of moving enquiries along, see fixing handoffs, and if the same information is typed in more than one place, stopping duplicate data entry is a good first step.