Which business process should you automate first? A simple scoring method

Automate the process that happens often, takes real time, follows clear rules, uses information already available in digital form, and causes little damage if a mistake slips through. For most small and mid-sized businesses, that means something like lead routing, data entry or follow-ups rather than a complex, judgment-heavy process.

The hard part is choosing. Most teams pick whatever annoyed them most recently. A quick scoring exercise gives you a far better answer.

Step 1: List your repetitive work

Ask each person on your team one question: "What did you do last week that a well-trained assistant could have done?" Write everything down without filtering it. You are looking for tasks, not job titles.

Common answers include:

• Copying enquiries from email into a CRM
• Typing invoice details into accounting software
• Sending reminders and chasing replies
• Building the same weekly report by hand
• Answering the same customer questions

Step 2: Score each task out of five

Give every task a score from 1 (low) to 5 (high) on these five factors:

• Frequency. How often does this happen? Daily beats monthly.
• Time. How many hours does it take in a typical week?
• Clear rules. Could you write down how to do it, including the exceptions?
• Clean inputs. Is the information already digital and reasonably consistent?
• Low risk. Could you catch and fix a mistake easily? Give a high score if yes.

Add up the five scores for a total out of 25. Anything above 18 is a strong candidate. Anything under 12 should wait.

Step 3: Check the payback

Scores tell you what is suitable. Payback tells you what is worth it. Use this simple sum:

1. Hours saved per week, multiplied by the loaded hourly cost of the person doing the work.
2. Multiply by 52 to get the yearly saving.
3. Compare that with the cost of building and running the automation.

As an illustration only: a task taking 8 hours a week at $25 an hour costs about $10,400 a year. An automation costing a fraction of that pays for itself quickly, even if it only handles most of the work rather than all of it.

Also count what the sum misses. Faster replies to leads and fewer errors in your records can matter more than the hours saved.

Choose a process that is boring and frequent. Boring work is where automation pays back fastest and fails most safely.

What not to automate first

• Broken processes. Automating a messy process only gives you faster mess. Fix the steps first.
• Rare tasks. If it happens twice a year, the setup will never pay back.
• High-stakes decisions. Keep a person in charge of pricing exceptions, legal matters and sensitive customer issues.
• Anything nobody can explain. If three people do it three different ways, agree on one way first.

Run a small pilot

Don't try to automate the whole process in one go. Take the first version through this path:

1. Pick one task and record how long it takes today.
2. Build the simplest version that handles the common cases.
3. Keep a person reviewing the output for the first few weeks.
4. Compare hours and error rates against your starting numbers.
5. Expand it, or stop, based on the result.

A good first project often leads naturally to the next one, such as automating follow-ups without losing the personal touch.

Should you use an automation tool or an AI agent?

If the steps never change, a standard workflow tool is cheaper and more predictable. If the task involves reading messy text, such as emails or documents, and deciding what to do with it, you need AI in the loop. Our guide to AI chatbots vs AI agents explains the difference.

Frequently asked questions

How do I know if a process is ready for automation?

Try writing the steps on one page, including what to do when something unusual happens. If you can, it is probably ready. If you can't, standardise it first.

Do I need to change my existing software?

Usually not. Most automation connects the tools you already use. Replacing software only makes sense if your current tools have no way of connecting to anything else.

How quickly will I see results?

A simple, well-chosen automation can show a measurable result within weeks. Larger projects take longer, which is why starting small is worthwhile.

Want a second opinion on your list? Book a free call and we'll help you pick the process with the best payback.