Poka-yoke (mistake-proofing)
Poka-yoke / Mistake-Proofing
Poka-yoke builds mistake-proofing into a process or equipment, so a human error either CANNOT be made or is detected the moment it happens — before it becomes a defect. The principle: people err; the process must catch it.
Use the method when
- →The same error type arises repeatedly despite instruction and attention
- →The root cause analysis has ended in “human error” — that is poka-yoke’s start signal
- →The error’s consequence is serious (safety, customer, expensive knock-on costs), even if the frequency is low
Don’t use it when
- →The error’s root cause is an unstable process (variation) — that is solved with stabilisation, not mistake-proofing
- →The solution becomes so cumbersome that it is bypassed — a poka-yoke people sneak around is worse than none
Before you start
The failure mode is precisely understood: which action, under which circumstances, with which consequence
Those who work in the process are part of the design — they know the bypass options best
How to do it — step by step
Describe the failure mode precisely
2-4 hoursWhat is done wrong (wrong part, wrong direction, forgotten step, wrong quantity)? When does it happen (start-up, switch, variant, stress)? The anatomy of the error decides the solution’s type — a forgetting requires something other than a mix-up.
Choose the level: prevent or detect
2-4 hoursLevel 1 (best): the error physically cannot be made — asymmetric plugs, guide pins, fixtures that only accept the correct part. Level 2: the error is detected instantly — a sensor, counter, weight check that stops the process or warns on the spot. Always aim for level 1 first; level 2 is the compromise.
Design simple and cheap
½-1 dayThe best poka-yokes are trivially simple: a guide bar, a stop, a template, a colour code, a counter. Brainstorm with those who do the work, and prefer the solution that can’t be bypassed over the one that is technically elegant.
Test against deliberate attempts to err
2-4 hoursActively try to make the error with the safeguard in place — in all the ways a pressured colleague might. If it can be bypassed under pressure, the design must be tightened; bypassability discovered after commissioning is in practice never discovered.
Standardise and record
2 hours + ongoingThe safeguard is written into the standard and the FMEA (detection/occurrence score is adjusted), and its function is checked in the daily/weekly routine — a defective sensor “temporarily” bypassed is a classic with a long history of harm.
The classic mistakes
How it goes wrong in practice, and this is where most implementations part ways.
Signs and instructions are called mistake-proofing
A warning note is not poka-yoke — it is a reminder, and reminders lose to busyness. The standard distinguishes hard: physically impossible or instantly detected; everything else is something else.
The safeguard is bypassed and stays bypassed
The sensor is disconnected in a pressured situation and never reconnected. A bypass must be visible (andon/board) and time-limited with an owner — otherwise the safeguard is an illusion in the statistics.
Overdesign
An 80,000-kroner solution where an 80-kroner guide bar would have caught the error. Complexity is itself a source of error — simplicity is a quality property.
The concepts behind it
Related standards
The standard is free. The anchoring is the craft.
The method only works when it becomes daily routines and leadership behaviour. The operations check shows in 4 minutes where your operations stand, and what should come first.