First pass yield: the quality number that hides nothing

First pass yield is the proportion of units that complete a process correctly the first time, without rework, repair or adjustment. It is deliberately unforgiving: a unit that was fixed and then passed does not count, because the effort of fixing it was real and the usual quality measures make it disappear.

What first pass yield measures that scrap rate does not

• Rework, which is invisible in a scrap figure because the unit eventually shipped.
• Adjustment and tweaking, which operators often do not record as rework at all.
• Re-tests and second attempts, where the first result was simply discarded.
• Hidden capacity loss: time spent on second attempts is capacity you paid for and did not get.
• Not customer returns, which belong to a different measure covering what happened after despatch.

Calculating it across several steps

For a single step it is units passing first time divided by units started. Across a sequence the honest version is the rolled figure — multiply the individual yields together — and the result is usually sobering. Five steps each running at ninety-five per cent give a rolled yield of about seventy-seven, which means roughly a quarter of everything needed intervention somewhere even though no single step looked troubled. That multiplication is the main argument for measuring per step rather than at the end.

Count the attempt, not the unit. The commonest way this measure gets quietly inflated is by recording a reworked unit as a pass once it finally conforms, which converts the one number designed to expose rework into another one that hides it.

What a high figure can conceal

1. Inspection that is not finding anything because it is not looking hard enough.
2. Rework done upstream of the measurement point, so the process being measured only ever receives good material.
3. Tolerances wide enough that nothing fails, which makes the number a statement about the specification rather than the process.
4. Operators correcting as they go without recording it, which is normal behaviour and invisible unless you ask.
5. A sample too small to show variation, where a single bad batch will later look like an anomaly rather than a pattern.

Ettex Sheets is enough to hold attempts and passes per step, which is all the rolled calculation needs and more than most small manufacturers currently record. Where the figure moves, the cause usually sits in the same place as the changeover and setup time, since first attempts after a change are where most of the losses concentrate.