Why factory software
stalls in Japan

Overseas industrial software companies describe the same thing to me. "The demo went extremely well. Then the calls stopped." Nothing was refused. No follow-up meeting is scheduled, and six months pass.

Filing that under "Japanese decision-making is slow" improves nothing. It is not slow. You have not identified where it stopped. There are four candidates.

1. The operating data may not belong to that factory

Many Japanese mid-sized manufacturers sit inside a larger company's supply chain. Under that structure, figures like equipment utilisation and defect rates are not merely internal improvement material. Read by the customer above them, they become price-negotiation material.

So the plant manager genuinely wants that visibility internally, and at the same time is wary of the data existing at all. "Let us make it visible" is a double-edged sentence for this person. The conflict is invisible to a foreign vendor.

2. The quality standard is not written down

Visual inspection is the most legible entry point for AI in a Japanese factory — ageing inspectors is one of the few problems management will state out loud. The interest is real.

The conversation then frequently stops at the request for training data, for a simple reason: the pass/fail criteria are not recorded anywhere. The veteran inspector knows a defect when they see one; no document defines why it is a defect. And they would rather not admit that to a prospective foreign supplier. So the answer stays "we have standards" and the data does not arrive.

This is not a weakness of Japanese manufacturing. It is a side effect of people spending careers on one site. At the point of introducing AI, though, it becomes decisive.

3. Maintenance is held by a long-standing local firm

Predictive maintenance also suits the market: equipment is old, the technicians who could hear a fault developing are retiring, and there are no successors.

But the maintenance work is usually outsourced to a local firm the factory has used for decades. Installing sensors and algorithms to detect anomalies also communicates that that firm's judgement is no longer considered sufficient.

That conversation is considerably harder than the budget conversation. Which is why the approval does not move.

4. Nobody wants to be first to sign for a foreign product

The last and largest wall. Japanese approval forms typically include a field for precedent — comparable deployments, other customers, adoption in the same industry. A new entrant has nothing to put there.

This is not refusal. It is nobody objecting and nobody proceeding. The foreign vendor reads it as lack of interest and withdraws — usually one step short.

And one more clock

The Japanese fiscal year begins in April. Where any public money is involved, a budget line that was not established before the year began means nothing happens that year. A proposal arriving in October was not rejected; it is roughly eighteen months early. Not knowing this means giving up at precisely the point where waiting was correct.

What to do instead

Where I am standing

I run Kohomadha Inc. from Tokyo and hold a public appointment as an overseas business support representative for a Japanese prefectural government. Because the practice works on both directions of the same border, I see the same wall from both sides.

In practical terms: I have developed and delivered an operations tool for a logistics company in Brazil, worked on digital transformation with a major Japanese retailer, and supported market development for a Swedish energy business and for a government project in Africa. Client names are withheld. That is deliberate.