- Why listing the work comes before tools
- How to inventory tasks line by line
- Spotting duplicate entry and single-person tasks
- Three ways to measure the effect
The short answer
The first step is not choosing a tool. It is writing down your current work, one task per line. Once you can see which tasks run on paper, which on Excel, and which on email or chat, the order of what to fix usually becomes obvious. Comparing systems can wait.
"DX" is a broad word, and it often gets read as "install software." This article is only about the work you do before that.
Why the first attempt stalls
If you start by comparing products, you eventually hit a question you cannot answer: where does an order come from, who receives it, and at what point is the quantity final? If you cannot describe that on one page, whatever you install will be configured around a process nobody has written down. The mismatch does not disappear — it gets filled in by hand, quietly, every month.
The second cause is scope. Trying to rebuild everything from order intake to invoicing at once pulls in more people and more decisions. A project that stalls halfway becomes a weight on the next one. Nobody wants to be asked, again, what happened to the last initiative.
Start with an inventory of the work
Paper or a spreadsheet is fine. One line per task, and only four columns. Keep the format plain — elaborate templates rarely get finished.
- The name of the task (for example, "receive the daily site reports")
- What it runs on (paper / Excel / email / chat / PDF / handwriting / spoken instruction)
- Who does it
- How often (daily / weekly / only at month end)
Cover one month, and only your own department to begin with. Trying to map the whole company at once means the list never arrives. Month-end tasks are the easiest to forget, so write the list with last month's calendar open next to you.
Write down what is actually done, as it is. Keep how things should be, and what you would like to change, on a separate sheet; mixing them puts steps that nobody really performs into the list and leads to wrong decisions later. Include work that is only ever asked for verbally, and the checks someone does on their own initiative, as lines of their own. Those lines rarely show up anywhere, yet things stop when they go missing.
When you are done, look for the same content appearing on more than one line. That is your next clue.
Look for duplicate entry and single-person tasks
Mark two kinds of rows.
Duplicate entry is any place where the same number or name is typed more than once. A typical shape: hours are written by hand on a site report, retyped into Excel at the office, and then copied a third time into the payroll sheet at month end. Every retyping is another place an error can enter.
Single-person tasks are the ones where the procedure exists only in somebody's head. The test is simple — go line by line and ask whether the task stops if that person is away for a week. If it stops, mark it. If one person collects a lot of marks, that is not a problem with the person. It means the procedure was never written down.
From the marked rows, pick the three with the highest frequency. Only three. Pick more and all of them end up half done.
What this looks like in practice
Say an interior fit-out company with eight staff finishes the inventory and gets this.
| Task | Runs on | Frequency |
|---|---|---|
| Site staff write hours on the daily report | Paper | Daily |
| Office retypes the report into Excel | Excel | Daily |
| Build the payroll sheet at month end | Excel | Monthly |
| File supplier invoices | Mix of paper and PDF | Monthly |
The first thing to fix is not the core business system. It is the retyping. If the hours become data at the moment site staff record them, the month-end sheet is just a summary of what is already there. And if dropping the paper report is not realistic yet, cutting the office retyping from two passes to one already changes the month. Invoice filing may not need software at all — agreeing on where files live and how they are named can be enough.
Japanese rules govern how business records may be stored electronically, and the guidance on retention periods and categories can change when the rules are amended. Check the National Tax Agency's current guidance before you settle on a filing method.
How to measure whether it worked
"It feels easier" does not survive contact with a busy month. Pick one of these three, whichever is easiest to capture, and compare before and after.
- Time: how many minutes per month the task takes
- Count: how many times the same data gets entered
- Errors: how many corrections or follow-up questions came up in a month
One month of measurement is enough. Precision matters less than having a "before" number at all. Without it, you have nothing to point at the next time you want to change something. You do not need a stopwatch; "about 10 minutes a time, 20 times a month" is precise enough.
Measure the work, not the people. Once the numbers are used to compare how fast individuals are, honest numbers stop coming. Show the totals to the people who do the work and check that they match their sense of it. If they do not, part of the procedure has been missed.
If a change did not help, resist the urge to make the number look better. A single line explaining why it did not fit is worth more the next time you decide something.
In short
DX is not the same thing as installing software. Write down one month of work, one line per task. Mark the duplicate entry and the tasks that depend on one person. Choose the three that happen most often, and start with the small ones. Then compare time, count, or errors before and after. If the project stalls partway through, you still keep the inventory — and that is the part you will reuse.
APOUL Clipboard
APOUL KINTAI
APOUL OCR Reader