How to price scrap and yield into a quote
Most quotes handle scrap with one line near the bottom. Add 2 percent, or 3 on a difficult part, applied to the finished cost. It is quick, everybody does it, and it is wrong in a way that is invisible until the job runs.
The problem is that it prices every loss identically. Scrap at the saw costs material. Scrap at final inspection costs material, 7 operations of machine time, the outside processing and the freight. Every figure below is illustrative and one routing carries through. It attaches to every line of a machined part build-up.
Why a flat scrap percentage prices the wrong thing
A percentage added at the end assumes losses happen at the end, and they do not. They happen throughout the routing, and the cost of each one is whatever had been spent on that piece at the moment it was condemned.
Take a 7 operation part. A piece lost at operation 10 has consumed material and 40 seconds of saw time. A piece lost at operation 70 has consumed material, 6 completed operations, a trip to the plater and back, and the inspection that caught it.
Those 2 events are not the same cost, so a single scrap rate cannot represent both. What the percentage actually does is average them, which is acceptable if your losses are evenly spread and misleading if they are concentrated somewhere, which they always are.
Cumulative yield across a routing
Yield multiplies down the routing rather than adding up. If each of 7 operations runs at 98 percent, the cumulative yield is 0.98 to the power of 7, which is 86.8 percent.
That is the number worth internalising. Seven operations that each feel excellent produce a routing where 13 pieces in every 100 started never ship. To deliver 1,000 good pieces you have to start 1,152.
Yield multiplies, it does not add.
The quoting consequence is a quantity multiplier before it is a cost adder. Starting 1,152 pieces means buying material for 1,152, running operation 10 on 1,152, and running operation 20 on whatever survived. Every operation upstream of the loss is paid for on the pieces that later die.
An estimator who adds 2 percent at the end has accounted for roughly a seventh of the real exposure on this routing. The gap does not appear in the quote. It appears in the job cost, months later, as an overrun nobody can explain.
Where in the routing the scrap actually hurts
Value accumulates along the routing, so the cost of a loss climbs with every operation it survives. On an illustrative part with $0.24 of material and 7 operations averaging $0.12 each, a piece lost after operation 10 costs $0.36 and a piece lost after operation 70 costs $1.08.
That is a 3 times difference for the identical event, which is why the useful question is not how much you scrap. It is where. A floor losing 3 percent at the saw has a cheap problem. A floor losing 1 percent at final inspection has an expensive one.
The estimating discipline is to carry yield per operation rather than per part. Most routings have 1 or 2 operations that account for nearly all of the loss, usually the tightest tolerance and the one nobody has fixtured properly, and those are the operations where the number belongs.
It also changes what you do about it. Moving inspection earlier does not reduce scrap, it reduces the cost of scrap, by condemning pieces before you spend another 5 operations on them. That is a routing decision an estimator can propose.
Find it early, pay for it once.
Outside processing deserves its own mention because it sits at the worst possible point. A part sent to a plater at operation 60 has absorbed almost all of its cost, and a rack that comes back wrong takes the whole value with it plus 2 weeks of schedule. The same reasoning that puts a machined part build-up on separate lines applies here, because the outside line carries a risk the inside lines do not.
First piece and setup scrap on short runs
Setup scrap is a fixed loss and it behaves nothing like running scrap. Dialling in an operation typically costs a few pieces regardless of lot size. On a run of 5,000 those pieces vanish into the noise. On a run of 50 they are a tenth of the order.
An illustrative 3 pieces of setup scrap per operation across 7 operations is 21 pieces. Against 5,000 that is 0.4 percent. Against 50 it is 42 percent, which means starting 71 pieces to ship 50, and no percentage-at-the-end approach will ever produce that number.
This is the single biggest reason a part quoted at production volume loses money on a prototype order. The same routing, the same tolerances, a completely different yield structure, and usually the same quote template.
Carry setup scrap as a piece count per operation and running scrap as a percentage. They are different quantities with different behaviour, and adding them into one figure destroys the information that makes short runs quotable.
Setup scrap is a count, not a rate.
There is a customer conversation buried in this. A buyer ordering 50 against a print quoted at 5,000 often expects the piece price to move a little. Showing that 21 pieces of the 71 you start are consumed proving the process explains the difference in a way that a higher number with no story behind it never will.
Where the material goes that never becomes a part
Yield loss and drop are 2 different leaks and both belong in the material line. Drop is the material you bought and never cut into a part, the remnant at the end of the bar and the skeleton of the sheet.
Yield loss is material you did cut, machined, and then condemned. On a bar fed part at 86.8 percent cumulative yield with 8 percent unusable remnant, the material you actually pay for per good piece is meaningfully above the weight of the part.
Keep them separate on the quote. Drop is a nesting and purchasing problem that better planning can improve. Yield is a process capability problem that needs engineering. Rolling them into one allowance means neither gets fixed because nobody can see which one moved.
The cycle time on every operation is also being paid on the scrapped pieces, which is the part estimators forget when they check their own arithmetic. Machine hours are consumed by everything you start, not by what you ship.
Putting it into the quote so it survives review
Quote the quantity you have to start, then price the operations against that quantity. It is the only method that gets both the material and the machine hours right, and it makes the arithmetic visible to anybody checking it.
State the yield assumption on the quote as a number. When the customer asks why 1,152 pieces of material appear on a 1,000 piece order, the answer is a sentence rather than an argument, and if their engineering improves the part the number can be revisited honestly.
The same structure carries into an assembly, where the compounding runs across levels instead of operations and a 3 level bill of materials turns 98 percent per level into 94.1 percent at the top.
Then measure it. A quoted yield that nobody ever compares against the actual becomes folklore within 2 years, and folklore is what the flat 2 percent was in the first place.