Poka-Yoke: Mistake-Proofing Work So Errors Can't Slip Through

When an error happens twice, the usual response is a reminder: an email, a sign, a stern word at the team meeting. Then it happens a third time. That isn't because people don't care. As a guide published by the Agency for Healthcare Research and Quality (AHRQ) puts it, people who make everyday errors "are not unmotivated or negligent," and they cannot eliminate those errors simply by telling themselves to do better.

Lean offers a different response: change the process so the error becomes hard or impossible to make, or so it's caught immediately. This is mistake-proofing, known in Japanese as poka-yoke. It's one of the most practical tools a small team can use, because good mistake-proofing devices are usually simple and cheap.

This guide explains what poka-yoke is, where it came from, the main approaches, and how to design one for a recurring error in your own work. For the same idea applied to keys, bills and household safety, see the companion guide Poka-Yoke at Home: Mistake-Proofing Everyday Routines.

What Poka-Yoke Means

The AHRQ guide, Mistake-Proofing the Design of Health Care Processes by John Grout, defines mistake-proofing as "the use of process or design features to prevent errors or the negative impact of errors." It describes poka-yoke as Japanese slang for "avoiding inadvertent errors." The Lean Enterprise Institute (LEI) breaks the term down the same way: "poka" means inadvertent error and "yoke" means prevention. LEI lists mistake-proofing, error-proofing and baka-yoke ("fool-proofing") as related terms.

According to the AHRQ guide, Shigeo Shingo formalized mistake-proofing as part of his contribution to Toyota's production system. The guide also notes that Japanese manufacturers formalized the approach in the 1960s and it was published in English in the 1980s, but that the idea is older. Its example is the elevator safety brake Elisha Otis demonstrated in New York in 1853.

Mistake-proofing fits a larger lean idea. LEI explains that lean producers build quality checks into the process itself, using operators and mistake-proofing devices, instead of relying on separate inspectors at the end. Its entry on jidoka, a pillar of the Toyota Production System, describes the goal as detecting abnormal conditions and stopping work immediately, so quality is built in at each step.

Slips, Mistakes and Knowledge in the World

The AHRQ guide draws on Donald Norman's distinction between two kinds of error. Mistakes come from deciding on the wrong intention. Slips happen when the intention is right but the action goes wrong. Because mistake-proofing requires knowing the correct action in advance, the guide notes, it works better on slips than on mistakes.

It also uses Norman's distinction between knowledge in the head (what people must remember) and knowledge in the world (cues built into the environment). Retraining, regular testing and efforts to keep people attentive all try to change what is in the head, and the guide observes that time erases those gains once the efforts stop. Mistake-proofing puts the knowledge in the world. It offers a rough test: if you can't take a picture of it in use, it probably isn't knowledge in the world.

Four Approaches

The AHRQ guide summarizes four approaches that it credits to an author named Tsuda:

1. Mistake prevention in the work environment. Reduce complexity and ambiguity. Its example is showing only one set of instructions at a time, so nobody reads the wrong page. Removing a step entirely removes every error that step could cause.
2. Mistake detection. Catch the error right after it happens, before it causes harm. Shingo called this informative inspection. The guide notes that self-checks, where people check their own work using a device, are preferred over checks by the next person in line because feedback is faster.
3. Mistake prevention. Check before the action, so the error can't occur. Shingo called this source inspection. The guide's example is the 3.5-inch diskette, which could only be inserted one way.
4. Preventing the influence of mistakes. Reduce the impact of an error. Examples include "undo" in software and the trash folder that holds deleted files, which the guide calls decoupling: separating the error from the point where it becomes irreversible.

How Strong Is Your Device?

The guide ranks the options. Prevention is usually preferred to detection. Among control methods, forced control, shutdown, warning and sensory alert are preferred in that order; the strongest devices need the least attention and the least discretion from users.

LEI describes two types: shutdown devices, such as a machine that won't start if a part is set incorrectly, and warning devices, which alert you with a buzzer or light. It says good poka-yoke are simple, long-lasting, low-maintenance, highly reliable, inexpensive and designed for the actual workplace.

Everyday examples from LEI and AHRQ include:

• A washing machine that won't start until the door is closed.
• A lawn mower that runs only while the safety bar is held.
• Spell-check in a word processor.
• A file cabinet that locks other drawers when one is open, so it won't tip over.

Designing One for Your Team

Imagine a hypothetical five-person print shop where jobs occasionally go out with the wrong quantity. A practical sequence:

1. Pick one recurring error with a real cost. Collect a few recent examples.
2. Go to where it happens and watch. When does it occur, and what is the person looking at in that moment?
3. Decide what to check. The AHRQ guide lists four "setting functions": physical attributes, the sequence of steps, grouping or counting, and making key information stand out against a noisy background. Wrong quantity is a counting problem.
4. Choose the strongest practical control. A warning works; a forced control is better. In the print shop, a job ticket could be designed so the quantity field must be filled in and is printed in large type, plus a counting tray marked at common quantities.
5. Test it, then watch for side effects.

Office examples follow the same pattern: required fields and drop-down lists instead of free text, templates with protected formula cells, color-coded folders for different clients, and a "send later" delay on outgoing email so a wrong message can be pulled back.

Watch for Side Effects

The AHRQ guide devotes a chapter to caveats. Its section titles include mistake-proofing the mistake-proofing, avoiding moving errors to another location, preventing devices from becoming too cumbersome and preventing worker detachment from the process. In practice, that means checking that the device doesn't just push the error downstream, isn't so annoying that people bypass it, and doesn't lull people into ignoring the work.

Key takeaways

• Mistake-proofing, or poka-yoke, changes the process or design so errors are prevented or caught at once, instead of relying on memory and reminders.
• AHRQ's guide credits Shigeo Shingo with formalizing it within Toyota's production system, though the idea is older.
• It works best on slips, where the right action is known but execution goes wrong.
• Prevention beats detection, and forced controls beat warnings; aim for the strongest practical device.
• Keep devices simple and cheap, test them, and watch that they don't create new errors or get bypassed.

Sources

• John Grout, Mistake-Proofing the Design of Health Care Processes, Chapter 1 (AHRQ Publication No. 07-0020, May 2007)
• John Grout, Mistake-Proofing the Design of Health Care Processes, contents (AHRQ, May 2007)
• Lean Enterprise Institute, Poka Yoke (Lean Lexicon)
• Lean Enterprise Institute, Error-Proofing (Lean Lexicon)
• Lean Enterprise Institute, Inspection (Lean Lexicon)
• Lean Enterprise Institute, Jidoka (Lean Lexicon)