Free tool . About 20 minutes

The Worth Automating Scorecard

Owners and operations leads at companies of 5 to 150 people deciding what to hand to software first.

Score any repeated task in two minutes and know whether to automate it now, fix the process first, or leave it alone.

Model prices just fell by half on the OpenAI side, so tasks that nearly made the cut last year may clear it now. Cheaper models do not make a messy process worth automating, though. This scorecard checks the task first and the technology second, which is the order that saves money.

How to use it

List five tasks. Write down five things someone on your team does every week that make them sigh. Do not filter yet. Quoting, chasing invoices, onboarding paperwork and weekly reports are the usual suspects.
Score each one. Answer the nine questions for each task. A yes earns the points shown, a no earns zero. Be strict: if you are not sure it is a yes, it is a no.
Read the band. Add up the points and find the band. Each band tells you the next step, and one of them is to leave the task alone.
Put a number on the winner. For the highest scorer, multiply hours a month by 12 and by what that person costs an hour. That is what the task costs you a year, and the ceiling on what fixing it is worth.

The scorecard

Tick every statement that is true for one task. Your score updates as you go. Print the page to score several tasks on paper.

Statement	Points
It happens at least once a week.	3
It takes someone more than two hours a month in total.	3
The steps are the same most times, and someone could write them down in ten lines.	3
The information arrives digitally, by email, form or spreadsheet, not on paper or by phone.	2
A mistake is cheap to catch, because nothing moves money or reaches a customer without a person seeing it.	2
The person who does it today agrees it is a problem.	2
You can name the number that would prove it worked, like hours a week or days to get paid.	2
It can run without access to the bank, payroll or anyone's passwords.	2
It is already written down somewhere, even roughly.	1

Score: 0 out of 20.

What the score means

Score	What to do
0 to 7	Leave it. Not worth a system. A written checklist or a better template will do more for less. Come back if it grows.
8 to 13	Fix the process first. Worth doing, not ready yet. Write the steps down and get the person doing it to agree on them. Then score it again.
14 to 20	Automate it now. Good candidate. Start with a version that drafts the work and lets a person approve it, then take the person out once the drafts are right.

A worked example

Chasing unpaid invoices at a 20 person company. Weekly, yes, 3. About three hours a month, yes, 3. Same steps each time, yes, 3. Arrives digitally from the accounting package, yes, 2. Mistakes cheap to catch? No, a reminder reaches the customer, so 0. The bookkeeper wants it gone, yes, 2. Proof number is days to get paid, yes, 2. Runs without bank access, yes, 2. Written down? No, 0. Total 17: automate it now, with a person approving each reminder for the first month. Three hours a month is 36 hours a year. At $35 an hour that is $1,260 a year, so a fix that costs more than that to build needs a second reason.

Or I can do this for you

If you want this done across the whole company rather than five tasks, that is The Automation Audit: two weeks, every repeated task written down, costed and ranked, with a payback number on every line. If one task scored 17 and the answer is obvious, you do not need an audit. Reply and I'll tell you what building it costs.

See The Automation Audit

Want the longer version? The Leak Finder is free on this site too.

From the AI Brief for 23 September 2026.