The Buffer Nobody Owned Until The Truck Was Late
A plant expected to shut down the lines that fill product because its film had not arrived.
The film arrived at 8:00
A plant expected to shut down the lines that fill product because its film had not arrived. The film landed at 8:00 on the day it was needed. Nothing stopped. No downtime code was written. No root cause was opened.
By the time the operations leader started reconstructing what happened, three explanations were already circulating, all of them from the supplier's side. The customer was supposed to be storing raw material on site under the agreement and was not. The supplier needed longer lead times than it was being given. An overseas shipping lane disruption had delayed the material. The plant's own team was leaning toward the simplest story available, which was that the supplier had failed.
What made the episode worth reading carefully is that the supplier is a highly capable one, with testing and validation resources most of its competitors do not have. When a strong supplier and a competent plant come within hours of stopping a line, incompetence is not the mechanism. Something structural is.
A clause is not a buffer
The protection in this case existed. It was negotiated, it was written down, and it was specific: raw material held at the plant. It had simply never been built.
That happens because a supply agreement and an operating practice have different owners. Sourcing negotiates the term. Legal papers it. The plant runs the schedule and places the orders. Nobody owns the step in between, which is converting a negotiated obligation into something the floor does and checks. So the agreement says stock is held on site, and no one holds it, and no one notices, because nothing in the daily rhythm of the plant asks the question.
The second half of the mechanism is why the gap persists. A line stop generates a downtime code, a root cause, and a meeting. A near stop generates a text message. The plant's institutional memory is assembled almost entirely from downtime records, which means the near miss contributes nothing to it. The exposure was real, the lesson was available, and the system recorded zero. Next quarter the same material will run thin again, and the organization will meet it with the same information it had the first time.
There is a people pattern layered on top of this that is worth naming, because it routes the investigation away from the fix. When material shows up late, the question that gets asked first is whose fault it is. That question always has an answer available from the supplier's side, because suppliers carry lead times, logistics, and upstream material constraints, and any of those will serve as a cause. The question that actually locates the repair is the uncomfortable one: what were we obligated to be doing in this agreement that we are not doing? Teams reach the first question in minutes and the second one rarely, because the first question has a defendant and the second one does not.
Convert the terms into a floor practice
The work here is small and it is almost never done.
Start by pulling the supply agreements for your highest volume materials and extracting only the operating obligations. Who holds stock, how much, where it sits, what lead time the quoted price assumes, what notice a specification or volume change requires. That list runs to a page per supplier, and in most plants it does not exist anywhere the people running the schedule can read it. The commercial terms get summarized and circulated. The operating terms stay in the document.
Then give each obligation an owner inside your own organization and a verification cadence. The person who confirms that consigned or on site stock physically exists should not be the person who negotiated it, for the same reason you do not let a buyer audit their own savings. Verification is a different job from agreement.
Then give the near miss a record. Define the trip wire in advance: when days of cover on a critical material falls below the level the agreement provides for, it gets logged with the same discipline as a downtime event, whether or not the line stopped. This is the piece that makes the loop close, because it converts an invisible event into a countable one.
Verify it a quarter later with a single question. How many cover breaches were logged, and how many of them became stops? If the answer is zero logged breaches and anyone on your team can still describe a scramble from memory, the trip wire is not wired and you have a reporting artifact rather than a control.
Keep the three states separate while you do this. A negotiated buffer, an implemented buffer, and a verified buffer are different things, and only the third one protects a schedule. Most plants are paying for the first and assuming the third.
What a well run inbound buffer reads like
Every critical material has one named internal owner and a stated days of cover. Current cover is readable from your own systems without calling the supplier. Consigned or on site stock is physically counted on a stated cadence, and the count is compared against the agreement rather than against the last count. The supplier's quoted lead time appears in the reorder parameters the planner actually uses, not only in the contract. A cover breach is logged inside the shift it occurs, and the count of breaches per quarter is a number the operations leader knows without looking it up.
Before you renegotiate price, find out what you already bought
The next time a critical material gets close, the useful move is not a supplier conversation about reliability. It is reading the agreement you already have and finding out which of its operating obligations are running and which were never switched on. That audit costs an afternoon, it requires no leverage, and it regularly turns up protection the plant has been paying for and never received.