What to Automate First in Your Business: A Simple Scoring Method

Automate the task that is frequent, follows the same steps every time, and costs you the most when it goes wrong or gets forgotten. To find it, list your repetitive tasks, score each one on four simple questions, and start with the highest score. That takes about 30 minutes and a notepad.

Most small businesses get this wrong by starting with whatever is most exciting, such as a chatbot or an AI assistant, instead of whatever is most costly. The method below is my own simple framework, not an industry standard. Adjust the weights to suit you.

Why does the order matter?

Your first automation sets the tone. If it saves real time and works reliably, people trust the next one. If it's fiddly, breaks, or touches something sensitive, the whole idea gets written off. So pick something that's likely to succeed, even if it isn't glamorous.

How do I score a task?

Write down every repetitive task you or your team do. Then score each from 1 to 5 on these four questions.

1. Frequency. How often does it happen? 1 is a few times a year. 5 is every day or many times a day.
2. Consistency. Does it follow the same steps almost every time? 1 means it's different every time and needs judgement. 5 means you could write the steps on one page.
3. Cost of the status quo. What does it cost now in hours, errors or missed money? 1 means barely noticeable. 5 means it eats real hours or loses you revenue.
4. Low risk if it goes wrong. If the automation makes a mistake, how bad is it? 1 means it could upset a customer or touch money or legal matters. 5 means a mistake is minor and easy to fix.

Add the four scores together. The maximum is 20. Start with the highest total, and be honest about question 4, because it's the one people inflate.

What does this look like in practice?

Say you run a small service business and you list these:

• Chasing unpaid invoices: frequent, same steps, costs you cash flow, and a polite reminder is low risk. It scores high.
• Replying to the same ten customer questions: frequent and consistent, but a wrong answer can cost you a customer. Medium, so keep a human check at first.
• Writing custom proposals: valuable, but different every time and high stakes. Low score. It's a poor first choice.

Those examples are illustrative, not data from any real business. Your list will differ, and that's the point.

What should I check before automating anything?

Run the top-scoring task through this checklist.

1. Can you write the steps down? If you can't explain it to a new hire in a page, a machine can't follow it yet. Write the steps first.
2. Is the process any good? Automating a messy process just makes the mess faster. Fix the obvious problems first.
3. Where does the data come from? If the inputs live in five places and half are missing, sort that out before anything else.
4. Where does a human stay in the loop? Decide in advance which steps are reviewed before they go out, particularly anything customer-facing or involving money.
5. How will you know it's working? Pick one number before you start, such as hours per week or days to get paid, and measure it before and after.

What are the common traps?

• Starting with the interesting task, not the costly one. A clever demo isn't a business case.
• Automating a rare task. If it happens twice a year, even a perfect automation saves very little.
• Skipping the human check. Early on, review what the automation produces. Loosen that once you trust it.
• Giving it access to more than it needs. Give any tool only the data and permissions required for the one task.
• Trying to automate the whole business at once. One task, working well, beats five half-working ones.

When this method doesn't apply

• If you have only a handful of tasks, you don't need a scoring sheet. Just pick the obvious pain.
• Some tasks score low but matter anyway. Compliance or safety work might need attention regardless of the numbers.
• The scores are rough. This is a way to start a conversation, not a precise calculation. Two tasks within a couple of points of each other are effectively tied, so pick the one your team is keenest on.
• Very small time savings may not justify the setup effort. If a task takes ten minutes a week, leave it alone.

If you want to see how this works for a specific function, we've written about what AI lead generation can and can't do. Our automation service covers sales, admin and finance workflows, and the consulting page explains how we help businesses work out where to start. If your best candidate turns out to be answering the phone or customer messages, see the AI receptionist and customer support chatbot pages.

Not sure which task to start with? Talk to us and we'll help you pick one.