Back-Solve the Number Before You Book It

The gap between 1,300 and 377

At a Midwest protein processor, an engineer spent four days on the trim line with a stopwatch. About 100 cuts, roughly 20 pounds each, an average cycle time of just under 53 seconds. That works out to 1,300 pounds per hour per trimmer while a trimmer is actively trimming. It is a clean number off a real sample.

Then he took the plant's own trim log, filtered it to flat cuts, and assumed 21 people on the line. All in, the same operation runs 377 pounds per hour per trimmer.

Both numbers are true. Neither one is a staffing plan.

The interesting thing is not the gap. The interesting thing is what the gap is made of. Divide 377 by 1,300 and the line is running at roughly 29 percent of its own touch rate. That is the implied condition sitting inside the all in number, and nobody had ever said it out loud. Knife sharpening does not account for it. Waiting does, and so does everything else that happens on a line when the work does not arrive evenly.

Here is the part worth copying. Rather than argue about where the missing 71 percent goes, the team picked a utilization assumption on purpose. Eighty percent. Not measured, not derived, chosen, stated, and owned, with the reasoning attached: nobody trims nonstop, and 80 percent is a defensible bar for work that is fundamentally hand paced. That single decision converted an observed rate into a headcount target of about 16 active trimmers, against a line that staffs 27 to keep 23 on it.

Every number implies a condition

The same week, in a completely different category, a buyer at an outdoor furniture maker opened a rotomolding quote that looked like a gift. About 141,000 pounds a year, and the quoted upcharges implied roughly $408,000 of annual savings, a weighted average of $2.96 per pound against the benchmark.

He did not book it. He solved backward.

The quote carried a 70 cent resin index marker, and several line items showed upcharges of only 88 cents per pound. But compound pricing in that material starts around $1.40 per pound. If the supplier is actually paying $1.40 and quoting as though the basis were 70 cents, then the implied processing cost is about 18 cents per pound. No rotomolder converts resin into a finished part for 18 cents a pound. A colleague ran the same check from the other direction: at the quoted rate, a 34 pound part prices out around $54, which is not a real number for that part.

So the number is arithmetic, the arithmetic is correct, and the condition it implies cannot exist. The response was not to discard the opportunity and not to bank it. It was to schedule a call to understand how the quote was built, while noting honestly that even if 70 to 90 cents comes out of the delta, something north of $200,000 may still be there. Speculative, and labeled speculative.

A third version of the same failure shows up in a refrigerated foods brand's carton award. One retail flat sleeve family looked like it had collapsed against its bid volume. The reason was structural: flats are not on any bill of material. The plant's bills call for glued cartons, and flats run as overflow when the automated line maxes out, planned as a roughly 30 percent split plus a safety stock number. The family was bid on a volume basis that does not exist in any system, and the supplier is now reopening minimums against it.

Three numbers, three categories, one mechanism. A rate measured while the work happens. A savings figure computed against an assumed input cost. A bid volume drawn from a basis with no bill of material behind it. Each is arithmetically sound, and each implies a condition that nobody stated until someone went looking.

Solve backward, then test on purpose

Solving backward is the cheap half. Do it on every number that is about to move money: state the condition the number assumes, then hand it to the person closest to the work and ask whether that condition happens. Eighteen cents a pound of processing. One hundred percent of a shift spent with a knife in motion. A carton volume with no bill of material under it. Conditions are falsifiable in a way that totals are not.

The expensive half is what you do next, and it is where most of these die.

Lead with the aggressive number and let the test walk it back. If the honest range is 16 to 19 trimmers, propose 16 and let validation push toward 19, then publish the range. Lead at 19 and you will never discover 16, because nobody tests upward into more work.

Make it one test, not a standing weekly argument. Define the break even metrics first: daily production made, labor hours not increased, work in front of the next operation held. A staffing change that gets relitigated every Monday dies on the first week production misses, and that week always comes.

Anchor the change to a natural break in the operation. That plant has a product changeover coming, and the plan is to start the new run at the target headcount rather than pull people off a line mid run. Same math, completely different conversation. It becomes a startup staffing decision instead of a reduction, and it converts the strongest objection, that all this work will be invalid once the product changes, into the trial itself.

What a well run number looks like

Every rate on the wall carries its utilization assumption and its sample size in the same field as the number. Every savings figure names the input cost it was computed against, and the person who wrote it can produce that input in under a minute. Bid volumes trace to a bill of material, or carry a written estimate with its logic attached and an owner's name on it, the way a 30 percent split plus safety stock is at least an assumption someone will defend. Staffing targets are stated as a range with a test date, break even metrics defined before the test starts, and a walk back path that does not require a new argument. Nobody is discovering an assumption from a supplier's price increase letter.

The condition, not the number

The trim rate was never wrong. The quote was never wrong either. What was missing in both was one sentence naming what would have to be true, and that sentence costs nothing until you skip it. Then it costs a staffing plan, or $408,000 that was never there.