Poka-Yoke (Mistake-Proofing)

Poka-yoke is a mistake-proofing device or method that either prevents an error from occurring or makes the error obvious at the moment it happens, before it becomes a defect that escapes to the next step or to the customer. The idea comes from the Toyota Production System (often paired with jidoka, or automation with a human touch): do not rely on vigilance alone when a simple design change can make the wrong action impossible.

Poka-yoke: an error-possible assembly versus a mistake-proofed fixture Left panel: a hinge leaf can be loaded either side up, so a hole is drilled in the wrong place. Right panel: a fixture pin and a unique pocket accept the part in only one orientation, so the drill cannot start if the part is flipped. Northriver hinge leaf: before and after mistake-proofing Before: error is possible Hinge leaf either side up Drill The blank looks almost symmetrical. A flipped part gets the hole 4 mm off. Defect found later, at assembly After: poka-yoke fixture Pocket + locating pin only one orientation fits Drill If the leaf is flipped, the pin will not seat. The start button stays locked out. The error cannot become a defect
A Northriver example. On the left, a nearly symmetrical hinge leaf can be loaded upside down and drilled off-location. On the right, a fixture pocket and locating pin accept only the correct orientation, and the drill will not start if the pin is not seated.

Error versus defect

An error is a departure from the intended method — a flipped part, a skipped torque check, a field left blank. A defect is the result that reaches a later process or a customer. Poka-yoke is aimed at the error, at the place the work is done. Inspection at the end of the line can still be useful, but it is late: the defective unit has already consumed time and material.

That is why mistake-proofing often appears as an action in an FMEA when a high-severity or high-occurrence failure mode is driven by a simple human or loading error, not by a poorly understood scientific cause.

Prevention and detection

Most practical devices fall into two families:

  • Prevention (control or shutdown): the wrong action cannot be completed. A keyed connector that will not mate if it is reversed, a fixture that will not close on a short part, or software that will not accept a ship date in the past. This is the stronger form.
  • Detection (warning): the error can still be made, but the process calls attention to it immediately — a light, a buzzer, a hard stop after a missed scan, or a checklist that blocks the next screen. Warnings work only if someone can and will respond.

A third, weaker habit is to add another inspector. That is not poka-yoke. It is more inspection, and it should be a last resort while you design a better device.

Common methods

  • Contact or shape methods use a physical or digital feature — a pin, a unique pocket, a keyed plug, a barcode that must match the work order — so the part or record can only go in one way.
  • Fixed-value (or counting) methods make sure the right number of actions happen: a tray with six pockets for a six-screw assembly, or a torque tool that will not release the fastener count until all six are recorded.
  • Motion-step (or sequence) methods watch the order of work: a sensor confirms that a safety cover closed before the press can cycle, or a form will not submit until required fields in order are complete.

The best device is usually the simplest one the people who do the work will actually use. A clever fixture that is slower than a workaround will be bypassed by the third shift.

Worked example: the Northriver drill fixture

Northriver Hardware drills a countersink in each hinge leaf. The leaf is almost symmetrical, and about once a day an operator loaded it flipped. The hole was four millimeters off. Assembly either forced the pin or scrapped the leaf. The failure was not a lack of training — the instruction already said “logo side up” — it was a design that allowed the error.

The team added a locating pin that engages a small existing notch on the logo side, and they wired the clamp-closed sensor to the drill start circuit. If the pin does not seat, the clamp will not show “closed” and the spindle will not start. Cycle time went up by almost nothing. The flipped-leaf defect went to zero in the following month. That is poka-yoke: the process, not the memory of the operator, carries the prevention.

How to choose a poka-yoke in a project

  1. Go to the place the error occurs (the gemba) and watch several cycles, including changeovers and recovering from a stop.
  2. Describe the error in one sentence and how it becomes a defect.
  3. Ask whether you can make the wrong action impossible (preferred) or immediately visible (next best).
  4. Involve the operators in the design, then try the device on a small scale — a kaizen PDCA cycle is a good frame.
  5. Put the new method into standard work and, if the risk was on an FMEA or control plan, update those documents so the device is maintained on purpose.

How this shows up on belt exams

Green and Black Belt items often ask you to recognize a poka-yoke in a short scenario, to distinguish prevention from detection, or to connect mistake-proofing to jidoka, source inspection, or FMEA actions. You may also be asked what poka-yoke is not — for example, a final-audit sampling plan. Related pages: FMEA, defects as waste, and the Body of Knowledge hub.

Frequently asked questions

Is poka-yoke the same as automation?

No. Automation can include mistake-proofing, and a fully manual fixture can be an excellent poka-yoke. The test is whether the method prevents or immediately surfaces an error, not whether a computer is involved.

Does poka-yoke replace training and standard work?

It does not. People still need to know the job, and the device has to live in the standard. Poka-yoke reduces the chance that a trained, well-intentioned person — especially under fatigue or interruption — can turn a simple error into a defect.

Where does the term come from?

It is Japanese. It is usually translated as “mistake-proofing” or “error-proofing.” Older English texts sometimes used “fool-proofing”; most modern Lean teaching avoids that phrasing because the point is to respect the worker and fix the process.

Related guides