Training does not fix a badly designed internal system

When a customer facing product has a difficult screen, it shows up as abandonment. Because it shows up, it gets a budget.

When an internal system has a difficult screen, nobody abandons it. Staff have no alternative. Because nothing shows up, nothing gets funded.

That asymmetry sits unaddressed in most companies for years.

The cost did not disappear

Nobody leaving does not mean nobody pays.

An expense system that takes ten minutes, used once a month by 300 people, costs 50 hours a month. 600 hours a year. Converted to salary, it is a real number.

That time appears in no department's budget. It spreads thinly across all of them and vanishes without becoming anybody's problem.

Japan's labour productivity is $60.10 per hour, 28th of 38 OECD countries. Uncounted time of this kind is one of the components of that ranking.

Training is the wrong instrument

Faced with a difficult system, most organisations respond with manuals and training sessions.

There is a structural problem with that response. Needing a manual means the screen requires explanation. A screen that requires explanation will always produce errors, because some proportion of people will not read the explanation.

Errors then add checking and rework steps. Those steps consume more of someone's time.

Add together the training cost, the manual maintenance and the rework, and the total often exceeds the cost of fixing the screen. But those three sit in separate budgets, so they are never added together.

Finding what to fix

You do not need to fix everything. There is a way to locate the expensive parts.

Count the operations that generate support requests. Your IT helpdesk log is a ranked list of usability failures. The top five usually account for most of the time.

Find where people have to re enter things. Screens that reject and return, applications where a defect surfaces after submission. These reliably consume time.

Look at where the manual is thickest. Page count is a direct proxy for screen complexity.

Measure approval waiting time. When an approver cannot tell what they are supposed to check, approvals sit. That queue is frequently an interface problem.

The fixes can be small

Improving an internal system rarely requires replacing it. In most cases this level of change is enough.

• Distinguish required fields from optional ones visually
• Show validation errors during entry rather than after submission
• Carry entered data forward to the next screen
• Order options by how often they are actually chosen
• Prefill with the previous value

None of these are large engineering projects. All of them remove time that is being spent daily.

As an entry point to DX

IPA's DX Trends 2025 identifies business architect as the role most lacking in Japan, the person who designs the join between the business and the technology.

Internal systems are a good place to develop that role. The users are in the building, so observation and validation are easy. Mistakes do not reach customers. And the effect is measurable in hours.

Learning progresses faster here than on a new external product. Companies that dismiss internal tooling are giving up that training ground as well.