Chapter three: The material, not the asset

The most quoted feature of the Toyota Production System is the one most often misunderstood. Just-in-time is remembered as an inventory strategy, a way of holding less stock and freeing up working capital. That was a consequence. It was not the reason.

Consider what a large buffer of incoming material actually does. A western plant in 1985 might hold six weeks of a given component. If a supplier's process drifted and started producing parts fractionally out of tolerance, those parts entered the warehouse, and the plant continued building with the good stock in front of them. The defect became visible six weeks later, by which point the supplier had produced six weeks of bad parts, the plant had no idea which batch was which, and nobody could reconstruct what had changed. Worse, some of the marginal parts had already gone into cars, and those cars were with customers.

Toyota held hours. A drift at the supplier showed up on the line the same day. The line stopped, the andon cord was pulled, and the problem was diagnosed while the evidence was still warm and while the people who had changed something could still remember changing it.

The inventory saving was real. The quality effect was the point. The buffer had not been protecting the plant from supply disruption. The buffer had been protecting the organisation from information about its own inputs.

Almost every enterprise data estate in 2026 is a warehouse in the 1985 sense. It holds years, not hours. It contains documents whose provenance nobody can state, whose currency nobody has checked, and whose fitness for any particular purpose has never been assessed, because until recently nothing consumed them at a rate that would have exposed the problem. Then a retrieval system was pointed at the estate, and the estate started answering questions.