Sample frameDay 327·Operations · Design

Good morning. Asking people to be careful isn't a control. It's a request you'll be making again next month.

Operations.

Poka-Yoke

5 min read·Apply by lunch

The question

When the same small mistake turned up for the third time — the blank field, the wrong file attached, the discount applied to the wrong line — did you change the step so it couldn't happen again, or send a message asking everyone to take more care? The message has already worn off.

The idea

Shigeo Shingo was an industrial engineer who spent years teaching and consulting inside Toyota's production system. In 1961, at Yamada Electric, he was shown a recurring fault: a small switch had two push buttons, each needing a spring underneath, and operators occasionally assembled one without its spring. The units reached the customer before anyone noticed. Shingo didn't ask for more training. He had the operator count two springs into a small dish before starting each unit — if a spring was still in the dish afterwards, the omission announced itself immediately, at the bench, for the price of a dish. He called such devices baka-yoke, fool-proofing, until a worker objected to being called a fool, and renamed them poka-yoke, mistake-proofing. The whole method sits in Zero Quality Control: Source Inspection and the Poka-Yoke System — 1985 in Japanese, 1986 in English — and it turns on one distinction: a mistake and a defect are not the same thing. Mistakes are inevitable, because attention is finite and people are people. A defect is a mistake that got through. So you either make the mistake impossible, or you catch it at the instant it happens rather than at an inspection later, once it has collected cost. The trap is filing every recurrence under discipline — a reminder, a retraining, a word with someone — which leaves the step exactly as easy to get wrong as it was this morning.

The framework

01
Two moves: make it impossible, or make it obvious.
Prevention beats detection wherever you can afford it — the three-pin plug that only fits one way, the diesel nozzle too wide for a petrol filler neck, the form that won't submit while a required field is empty. Where the physics or the software won't allow prevention, fall back to detection at the moment of the error rather than at the end of the line: Shingo's dish with a spring still in it, a total that has to agree with a count, a confirmation screen that shows what's about to be sent instead of asking whether you're sure. Either beats hearing it from the customer.
02
Shingo's three ways to catch a mistake.
The first is contact: shape, size or format does the checking, so the wrong thing physically won't go in. The second is fixed value: something must occur a set number of times, and a wrong count is the alarm — two springs in the dish, six bolts in the tray, four attachments on the email. The third is sequence: the next step refuses to begin until the previous one is confirmed, which is why a well-built system won't let an invoice go out before the address has been verified. Most office and service errors yield to one of the three, and usually to the second.
03
Why is this step easy to get wrong?
Mistake-proofing starts from a specific error mode, not a general area of concern. Invoicing errors gives you nothing to design against; the delivery address gets copied from the previous order at line three of the form gives you a device in about a minute. Shingo called the discipline source inspection — go after the condition that produces the error, not the defect that the error produces. Pull the last month of mistakes, write each as a single sentence naming the exact moment it occurred, and you'll usually find three or four moments accounting for most of them.
04
The device should cost less than the mistake it prevents.
Shingo's devices were deliberately cheap — a dish, a jig, a coloured mark, a template with the risky fields pre-filled. If your proposed control is a new system and a fortnight of build, you've probably picked an error that doesn't earn it, or you're solving the general case when the specific one would do. Two failure modes to watch for. A control that people can bypass will be bypassed in the first busy week, so design it out. And a warning added to a screen already crowded with warnings buys nothing, because nobody reads the fifth alert.

Members only · 4 principles + template + AI mentor

364 more frameworks are waiting.

Poka-Yoke is Day 1 of 365. One framework every morning for a year — across strategy, sales, negotiation, leadership, and more.

$1 /day

Billed $365/yr · cancel any time

365 frameworks

12 topics · templates · AI mentor

Reserve your place — $365/yr →

One payment a year · the whole library · no auto-renewal.

More Operations frameworks

All 24 operations frameworks →