What one repetitive workflow actually costs you

Ask a business owner which process wastes the most time and they will answer in about two seconds. Ask them what it costs and the room goes quiet. That gap is the whole reason most automation never gets bought: annoyance does not survive a budget meeting, but a number does.

Here is the arithmetic. It takes four inputs and about five minutes.

The four numbers

1. Runs per month. How many times the process happens. Count it for a week and multiply — do not estimate, because people underestimate this by roughly half.
2. People per run. How many humans touch it. Not how many are responsible; how many actually open something.
3. Minutes per pass. Time one real pass, including the context switch before it and the checking after it.
4. Loaded hourly cost. Salary plus everything else, divided by working hours. For most roles this is 1.3x the number people quote.

Multiply them. That is the monthly cost of doing it by hand.

A process that runs 80 times a month, touches two people, takes 11 minutes a pass, at $28 an hour loaded, costs $821 a month. That is $9,850 a year to move information between two systems that both have APIs.

The part people get wrong

The mistake is counting only the doing. Three costs sit outside the stopwatch and they are usually larger than the one inside it:

• Error cost. How often does it go wrong, and what does fixing it take? A 4% error rate on a client-facing process is not a 4% cost.
• Delay cost. A lead that waits two days is not the same lead. A report that arrives late is not the same report. Sometimes this dwarfs the labour.
• Ceiling cost. The work the person would be doing instead. This is the real one, and it never appears in a spreadsheet.

What the number is for

It is not for justifying my invoice. It is for deciding whether to bother at all.

If the manual version costs $200 a month, do not automate it. Delete it, or leave it alone, and go find the one that costs $3,000. I turn away more of these than I take, because a fixed-price build against a $200 problem is a bad trade for both of us and I would rather say so on the call.

If the number comes back above roughly $1,000 a month, the payback period on a single-workflow build is usually two to three months. At that point it stops being a productivity conversation and becomes an obvious one.

Do this before you talk to anyone

Run the arithmetic on your three worst processes before you take a single sales call — mine included. You will find one of three things: the problem is smaller than it felt, the problem is exactly what you thought, or the problem is enormous and nobody had ever multiplied it out.

The third one happens more than you would expect.