Automate first the task that happens often, follows rules a person could write down, is easy to undo when it goes wrong, and already has its data in one place. Score your candidates on those four tests before spending anything. The task that annoys people most is often not the one that pays back first.
Four tests for a good first task
Most businesses pick their first automation by frustration. A better way is to score every repetitive task against four tests, and start with the one that passes all four.
A task that fails one test can still be automated later. It just should not be the first, because the first project sets how much the team trusts everything after it.
- Volume. It happens many times a week, so small savings add up and problems show quickly.
- Rules. A person could write down how it is done. If the answer is usually "it depends", it is not ready.
- Reversible. A mistake can be caught and undone before it reaches a customer or a bank account.
- Data in reach. The inputs are already digital and in one place: a form, an inbox, a spreadsheet, a CRM.
How common tasks score
These are the tasks we are asked about most. The verdict is about where to start, not about what can never be automated.
| Task | Volume | Rules | Reversible | Verdict |
|---|---|---|---|---|
| Copying enquiries from a form into a CRM | High | Clear | Yes | Strong first candidate |
| Appointment and payment reminders | High | Clear | Yes | Strong first candidate |
| Sorting incoming enquiries by type | High | Mostly | Yes | Good, with a person checking edge cases |
| Drafting replies to common questions | Medium | Partly | Yes, a person sends | Good as a draft, not an automatic send |
| Approving refunds or discounts | Low | Partly | No | Leave with a person |
| Paying suppliers | Medium | Clear | No | Automate the paperwork, not the payment |
Put a number on it before you build
Count the hours the task takes, at a fully loaded hourly cost, then subtract the share a person will still do afterwards for exceptions and checking. Add what it costs to build and to run. Our automation ROI calculator does this arithmetic and refuses to assume the task disappears completely.
As an illustration: three people spending six hours a week each on a task, at GHS 60 an hour, with a quarter of the work still needing a person, comes to 621 hours a year handed back. With an assumed build cost of GHS 18,000 and GHS 600 a month to run, it pays back in 8 months. Those inputs are assumptions for the example, not a quote. Run your own.
Run the first one small
The first automation should be boring on purpose. Its job is to prove the method, not to transform the business.
- One task, one owner, and a written definition of what a good result looks like.
- Run it beside the existing process first and compare the two before switching over.
- Keep a way for a person to take over at any point, and make it obvious.
- Measure time saved, errors and manual corrections for a month before choosing the second task.
The best first automation is rarely the most ambitious one. Pick the task that is frequent, rule based, reversible and already digital, put a number on it, run it small, and let the results decide what comes next.
What is the easiest thing to automate in a small business?
Usually moving information that already exists into another system: enquiries from a website form into a CRM or spreadsheet, or reminders sent from a booking calendar. The rules are clear, the volume is steady, and mistakes are easy to spot and fix.
Should a business start with an AI agent?
Rarely. Most first tasks are covered by fixed rules, which are cheaper, more predictable and easier to test. An AI model earns its place where the input varies, such as reading free text to sort an enquiry, and even then a person should check the output at first.
How do I know if automation is worth it?
Count the hours the task takes, subtract the share a person will still do, and compare the saving with the cost to build and run it. If the saving does not cover the running cost, it never pays back, however clever the automation is.
Turn the research into a working plan.
Bring the evidence, constraints, and next decision. We will help you turn it into a clear technical brief.