No Protocol, No Spec Change, No Savings

The film that nobody could approve

A multi-plant protein processor had a flexible film program that looked like an easy win on paper. The current film had been qualified years earlier by an internal engineer who had saved the company real money doing it. The category was up for renewal. A supplier extension was already in place through early 2027, which bought time. Everything about it read as a straightforward sourcing exercise: go to market, take the savings, move on.

It stalled, and not on price. The engineer who qualified the incumbent film was not an ally for change, and he was not being unreasonable. Nobody could tell him what the current film actually did. There was no scrap rate on it. There was no leaker rate. There was no downtime attribution. When the question is "will this new film run as well as the one we have," and nobody can state how well the one we have runs, the honest answer is that the question cannot be answered.

Two other things were happening at the same time. The savings target kept moving. Each conversation produced a bigger number, from around $400K to north of half a million, with no new data behind the growth, which meant the program was accumulating an expectation it was unlikely to meet. And the risk was asymmetric in a way everyone on the floor understood: if there is a film issue three years from now, it gets blamed on the change, regardless of cause. A qualified incumbent is not just a spec, it is somebody's professional record.

A spec is an input to every step downstream

Procurement treats a material spec as a line item with a price. The floor treats it as an input to every process step that follows. Those are different objects, and the gap between them is where packaging savings go to die.

Change a film and you have changed sealing behavior, machine speed, tension, giveaway, and reject rate, all at once, in ways that show up in different departments on different days. The incoming variance propagates. That is not an argument against changing it. It is the reason the change needs a test rather than an opinion.

The failure mode is subtler than resistance. It is testing without a definition of done. Run trials, look at the results, ask the room how everyone feels, run more trials. You are never going to feel good. There is always one more line, one more SKU, one more season to check, and a validation program with no acceptance criteria will consume SME time, subcontractor cost, travel, and production slots indefinitely and still end in a judgment call made by whoever is most senior in the room. That is the expensive version of no.

The second-order effect is worse than the wasted trials. Once a category acquires a reputation for endless validation, the next spec change does not get proposed. The organization learns that material change is a swamp, and it stops looking at the largest controllable cost in the plant.

Write the acceptance test before you run the trial

The fix in that engagement was to stop the sourcing motion and build the protocol first, as a client-owned document rather than a consultant deliverable or a supplier deliverable. That ownership detail matters more than it sounds. A supplier's test protocol will be shaped to what that supplier's material passes. A protocol the plant owns applies to the incumbent and every challenger equally, and it survives the project.

The document was built around two questions, and only two.

Does it physically run? Line speed against the incumbent, downtime, and changeover behavior, pulled from the automated line-monitoring system rather than manual tracking. Passive data is the point. When operators know a trial is being watched by hand, you measure the attention, not the film. When the data is already being collected invisibly, you measure the film.

Does it hold quality? Leak testing and the existing quality checks, run at the same frequency, with the pass threshold written down before the trial starts.

Everything else that usually clogs these programs, supplier relationship, contract timing, who qualified what in which year, was pushed out of the test and back into the commercial track where it belongs.

That is what unlocked the category. The strategy that came out the other side was to go to market like for like, so the incumbent spec sets a clean price baseline, and to develop an alternative spec with each supplier as a gated second step: the alt spec has to show the prize before anyone spends validation money on it, and it has to clear the written protocol before it goes into a plant. Same category, same suppliers, same team. The difference is that a no is now cheap and fast, and a yes is defensible three years later when somebody goes looking for who to blame.

Note the sequence. The protocol came before the trial calendar, and the trial calendar came before the market event. If you run it in the other order you will be negotiating a price for a material you have no way to accept.

What a well-run spec change looks like

Baseline scrap, downtime, and leak-test rates for the current material are on a dashboard before anyone talks to a supplier. The acceptance protocol is one document, owned by the plant, that names the metrics, the trial length, and the pass or fail threshold for each. Trials have dates on a calendar, not a standing invitation. Verdicts come from line-monitoring data at the same collection frequency used in normal production. Alternate specs get a savings gate before they get a trial slot, and any spec that fails the protocol is closed out in writing so the next team does not re-litigate it. Under that regime, a spec change is a two-question decision with a date on it.

Closing

The engineer holding the line on that film was doing his job correctly. He had no way to measure the risk he was being asked to accept, so he priced it at infinity. Build the measurement and the objection dissolves, because it was never really about the film.