Every owner we sit down with has a list. Somewhere between the napkin and the project-management tool, there are five or ten things that “should really be automated someday.” The list is usually wrong — not because the pain isn’t real, but because it ranks workflows by annoyance instead of payback.
Annoyance is loud. Payback is quiet. The workflow that makes everyone groan in the Monday meeting is rarely the one bleeding the most money. This article is the method we use in our free Operations Audit to find the difference, and you can run most of it yourself with a notepad and twenty minutes.
Start with hours, not features
Skip the software ideas entirely for now. Make a list of recurring work — anything that happens weekly or more often — and put two numbers next to each item: how many hours a month it consumes, and the loaded hourly cost of the people doing it.
Be honest about the hours. “Entering orders” is not 45 minutes a day; it’s 45 minutes a day plus the re-entry when the spreadsheet and the ERP disagree, plus the Friday afternoon someone spends reconciling the two. Count the cleanup, not just the task.
Most businesses we audit find that two or three line items account for the bulk of the recoverable time. A manufacturer we worked with assumed scheduling was their biggest problem because it caused the most arguments. The numbers said order intake was costing three times as much — about 320 hours a month. Arguments are loud. Data entry is quiet.
Then add the two costs nobody writes down
Hours are the visible cost. Two others usually matter more:
- Error cost. What happens when the manual step goes wrong? A mistyped order quantity, a missed onboarding email, a stale price on a quote. Multiply the frequency by the cleanup cost — including the customer goodwill, which is real money even if it never hits a ledger line.
- Wait-time cost. How long does work sit between steps? If a quote takes four days because it crosses three desks, you are losing deals to whoever answers in one. Automation’s biggest wins are often latency wins, not labor wins.
A workflow with modest hours but brutal error or wait costs can out-rank everything else on the list. This is where gut feel fails and a few minutes of arithmetic pays for itself.
The three filters that disqualify a workflow
Plenty of painful workflows should stay manual. We walk away from builds when any of these is true:
- The process changes every time. If the team can’t describe the workflow’s happy path in five sentences, software will just freeze the chaos in place. Standardize first, automate second.
- The volume doesn’t justify it. A task that takes two hours a month is a bad target no matter how annoying it is. The build would never pay itself back.
- The judgment is the work. Some steps look like data shuffling but are actually a human making a call — approving credit, pricing an unusual job. You can automate everything around that decision and put it in front of the right person faster, but the decision itself should stay human. This is exactly where we draw the line in our AI agent builds: agents prepare the work, people approve it.
Score it like a buyer, not a builder
For each surviving workflow, you now have monthly hours, error cost, and wait cost. Add them up, annualize, and compare that number to a realistic build cost. A useful rule of thumb from our own quoting: a focused workflow build pays for itself in under six months or we tell the client not to do it.
That payback math is the whole reason we quote fixed prices with delivery dates instead of hourly estimates. You can’t make a payback decision against a meter that’s still running. If you want to pressure-test your own numbers, the ROI calculator on our pricing page uses the same logic.
What this looks like in practice
Run the method on a real example. A professional-services firm tracks client onboarding in email and a spreadsheet:
- Hours: about 6 per new client, 8 clients a month — 48 hours.
- Error cost: one missed document request a month, which delays billing by a week.
- Wait cost: clients sit idle for 3 to 5 days between signing and kickoff, and two a year quietly churn during that gap.
Annoyance ranking would have put “the partner expense reports” first. Payback ranking puts onboarding first by a mile — and that’s the build that cut onboarding from days to under an hour in one of our case examples.
Do the twenty minutes
You don’t need a consultant to run this. List the recurring work, attach hours and costs, apply the three filters, rank by payback. If the top item survives scrutiny, you have your first automation project — and a number that tells you what it’s worth paying to fix.
And if you’d rather have a second set of eyes, that’s literally what our free 20-minute Operations Audit is: we run this method on your business, hand you the ranked list with a fixed price and a delivery date for the top item, and tell you straight if software isn’t the answer.
