Measure first, then automate
Most requests arrive before anyone knows how many hours are actually being lost. Here is how to measure it in one afternoon.
Tim Neuraltan · 22 July 2026 · 6 min read
Most requests reach us in a similar form: "Build us something so the team stops entering data by hand." Our first question is always the same, and it usually pauses the conversation: how many hours does that manual entry cost you right now?
Almost nobody knows.
Why the number matters
Without it, you cannot tell whether a system costing tens of millions of rupiah pays for itself in three months or three years. You also cannot tell which part deserves to be built first.
Measuring it in one afternoon
Take one repetitive task. Ask the person who does it to record three things for a week:
- How often the task occurs
- How many minutes each occurrence takes
- How often it must be redone because of an error
Multiply: occurrences per month × minutes each ÷ 60 = hours per month. Multiply by that person's hourly cost. That is what is currently being lost.
A real example from a logistics client: 1,400 shipping documents a month, 4 minutes each to reconcile against contract rates, plus 5% needing rework. That came to 98 hours a month — one full-time person doing nothing but copying numbers.
Where people get it wrong
Counting only salary. Lost time delays other things too: invoices go out late, payments come in late. Count the delay.
Forgetting the cost of errors. A mis-rated invoice is not just corrected — there is a phone call, an apology, sometimes a discount.
Measuring a quiet week. Measure the busiest week. Systems are built to hold the peak, not the average.
If the number is small
If it works out to two hours a month, do not automate it. We will tell you so, even though it means no work for us. Building a system for something that rarely happens only adds one more thing to maintain.