You Can't Bank a Feeling

We Did It Until It Felt Good

A protein co-packer wanted help locking in savings on a flexible-film conversion. The pitch already had a hero. An engineer there had qualified the film a cycle earlier and come back with a big number, somewhere around twenty percent, and the story had hardened into fact. The film was perfect. The savings were booked in everyone's head. So when we started asking how the target had been set and how it would be verified, the room got skittish, and the goal line started to move.

The tell came in one sentence. Asked how the original qualification had landed on its target, the answer was, "we just kind of did it until we felt good." There it was. No pre-change readings. No measurement method. No date. The savings were real to the person who found them and invisible to everyone else, because the only record was a feeling. As the internal note put it bluntly: we don't have good baseline data. That is the whole problem. You cannot defend a number whose starting point was never written down.

The Reference Point Is the Asset

A baseline is not paperwork. It is the asset every later number leans on. Get it wrong and you do not lose one figure, you lose the reference frame for all of them. The twenty percent is not just unverifiable on its own. It silently sets the anchor for the next conversation, the next phase, the gain-share split, and the case for bringing subcontractors back onto the floor. When the start point is a feeling, the error does not stay contained. It compounds forward into decisions that cost real money.

This is why goalposts drift. People assume the target moved because the client got difficult. Usually the target moved because there was never a post in the ground. If nobody recorded where the line ran before the change, then "better" is an opinion, and opinions slide. The plant defends a number it cannot reproduce, sourcing quotes a number the plant will not stand behind, and both sides slowly stop trusting each other's math. The relationship frays not over the work but over the scoreboard nobody built.

The same disease shows up wherever the measurement is left to the party being measured. On another engagement, a facility manager was reporting strong efficiency while quietly withholding the raw output from the line-monitoring system. Self-reported efficiency is the corporate cousin of "we did it until it felt good." The number sounds like data. It is a claim wearing the costume of data. Until you can pull the readings yourself, you do not have a baseline, you have a press release.

Set the Post Before You Swing

The fix is unglamorous and it works. Before any change, capture the starting state in a form a stranger could audit: the actual readings, the method used to take them, the date, and the conditions. Independent capture matters as much as the number. If the line reports its own efficiency, you are not measuring the line, you are measuring its mood. Pull the data from the system, not from the operator's summary of the system.

Then refuse to inherit someone else's feeling as fact. When that protein co-packer's twenty percent could not be traced to a measured start, the right move was not to argue about the film. It was to hand back a defined baseline and framework and say the next phase does not move until validation protocols are finished. That is not bureaucracy. It is the only thing that lets you eventually bank a real number instead of relitigating a remembered one.

The discipline also tells you when to walk. A separate prospect wanted a fifteen-location, $225M footprint diligence delivered in three weeks. The honest read was that no one could produce a number worth acting on in that window, across that many sites, with that little leverage. A figure you cannot baseline inside the timeline is a figure you should not sign your name to. Declining was the disciplined answer, not the timid one. The lighter alternative we offered, a market scan with a clear method and a clear limit, was worth more than a confident three-week guess, precisely because it did not pretend to a precision it could not measure.

If you run a plant or buy for one, the diagnostic is fast. Pick any savings or efficiency number currently being defended and ask the person who owns it to show you the before. Not the after, not the projection, the before: the readings, the method, the date. If they can produce it, you can build on it. If they reach for "we knew it was better," you have just found a feeling that has been quietly setting your reference frame, and every decision downstream of it is exposed.

The Lens Flip

The savings were never the problem. The missing baseline was. A twenty percent gain you cannot show me is not a smaller gain, it is a zero, because the only thing a decision table can spend is a number with a measured start. Set the post before you swing, or accept that you are not optimizing a process, you are negotiating a memory.