One Number Per Supplier: The Plan to Find Bottlenecks Without Full Disclosure

A readiness survey of more than one hundred pages, taking three months for a single round, still could not tell an aircraft manufacturer where its network would break. That failure sits at the center of a new four-part series on multi-tier supply chains by Brian (Yoshiatsu) Kawabata, a veteran of Honda, Boeing and Intel, whose opening installment frames a problem every shipper with deep supplier networks will recognize: the constraint that matters most is the one nobody can see, and nobody will show you.

The series, part of a sequence publishing on Nov. 10, 17 and 24, argues that supply chain directors should hunt for bottlenecks during calm periods, before a production ramp-up exposes the weakest node. Kawabata knows the alternative intimately. At Boeing, he writes, he saw the state the OEM and its peers call recovery — the emergency posture a network drops into once backlog piles at a supplier and a stoppage looms. Boeing runs a standing supplier recovery team that sends its own engineers into struggling suppliers to raise output. It works, he concedes, but it is by definition a response after the damage is done.

What can actually be controlled?

Kawabata draws a hard line between two kinds of risk. Disruption from disaster or accident falls across every firm at roughly equal probability; you insure against it, site plants away from hazards, and accept that rivals face the same exposure. What he calls inexcusable is leaving unaddressed the controllable risk — the weakest process in the network failing to keep pace as production rates rise.

His image is a marching band of members with very different fitness, gradually asked to march faster. Sooner or later the least fit member falls behind. A planned ramp-up, he argues, is precisely the moment to lighten that member's load before the parade sets off, because two things are known in advance: the ramp is coming, and the weak point can be found if you search for it early.

The competitive logic is blunt. Managing controllable risk will not prevent every disruption, but when a rival stumbles on a risk you handled — when they could not ramp and left customers waiting — you take their share. He invokes the old joke about the bear: you do not need to outrun the bear, only the person beside you.

Why visibility depends on network depth

Kawabata has worked the buyer-supplier relationship from both sides, and the need for visibility was opposite at each. At Intel, the upstream network is small, shallow and narrow; the responsible managers knew their suppliers' situations — and those of the equipment makers one tier further up — in considerable detail. A network-wide bottleneck-hunting tool had little role there.

Boeing was the reverse. The number of suppliers behind a single aircraft is of another order, and in automotive the count can run to thousands of firms. Beyond Tier 1, visibility grows dim. His conclusion: visibility is not universally necessary. It becomes necessary only once a network passes a certain depth and breadth — and complex products like cars and aircraft passed that threshold long ago.

The hundred-page questionnaire

The series' most concrete cautionary tale is the Rate Readiness report. At the aircraft manufacturer, the OEM imposed it on Tier 1 heavy-industry suppliers: a survey well over one hundred pages on preparedness for a future rate increase, mandatory for every firm. Each department filled in its dozens of pages, sales stitched the answers together, and a single round consumed three months. It covered capital investment and much else — a heavy burden that, Kawabata notes pointedly, still did not stop supply disruptions from surfacing.

Two structural flaws stand out. The survey may not carry through to Tier 2 and beyond, where the deeper constraint may sit. And demanding that every item be answered by every Tier 1 firm lowers the quality of the whole. He offers a teacher's analogy: when learning a language, would you rather memorize the entire dictionary or study only the grammar that matters? The burden, he argues, should be narrowed to exactly what locating the constraint requires.

Why suppliers keep the door closed

The heart of the article is a confession one supplier made to Kawabata after years of building trust. Almost no supplier ever accepted his proposal to find the bottleneck process together in calm times; only in crisis, with backlog piling and a stoppage looming, would they grudgingly allow a visit and let the OEM measure process capacity. Then one supplier said aloud what firms rarely admit: 'You can tell me to hand over the internal process information. But at next year's contract renewal you'll use it to negotiate the price down, won't you? So how could I hand it over easily?'

Kawabata calls that entirely reasonable. A one-sided demand for disclosure is, in his words, nothing but exploitation — and the data could leak to a rival who undercuts the supplier and takes the business. Structurally, an OEM simply demanding information cannot work: ask every firm equally, and one stays quiet hoping not to stand out while another fears that honest answers will make it the lone loser. Cooperation requires one of three things, he writes — overwhelming leverage, a benefit to the firm for answering, or grounds strong enough to justify the request.

One number instead of full disclosure

The proposal the series will develop is an information-sharing platform that asks each firm for a single number, and from those alone locates the network-wide constraint in peacetime. Everyone learns who carries the weight of the whole, and no firm loses by disclosing. The key, Kawabata teases, has been in plain sight for over half a century: two old tools that grew up separately in project management and production management, rarely brought together — and which, he argues, are saying the same thing.

His credentials underpin the claim. Kawabata began at Honda, earned an MBA at Toulouse Business School in 2008, and completed a PhD in computer science at Tokyo University of Agriculture and Technology in June 2026, with a dissertation on detecting multi-tier network constraints from minimal shared information; a related paper appears in Information Technology and Management (Springer, 2026). Part 2, due Nov. 10, will lay out whether those two legacy tools can genuinely compress bottleneck detection into a number per firm — a question with direct consequences for how OEMs and their suppliers structure the next round of capacity talks.