Ask why three times, not five

The original is worth reading, because almost nobody has. Taiichi Ohno, in Toyota Production System, under a heading that says exactly what it does:

Why did the machine stop? There was an overload and the fuse blew. Why was there an overload? The bearing was not sufficiently lubricated. Why was it not lubricated sufficiently? The lubrication pump was not pumping sufficiently. Why was it not pumping sufficiently? The shaft of the pump was worn and rattling. Why was the shaft worn out? There was no strainer attached and metal scrap got in.

It is a beautiful piece of reasoning and it works completely. Ohno’s own point: without it, “one might simply replace the fuse or the pump shaft. In that case, the problem would recur within a few months.”

Notice what he is asking, though. He is asking a machine.

A machine has one chain. A business need does not

The lubrication pump has exactly one reason for not pumping. Follow it down and you arrive somewhere real, because there is only one somewhere to arrive at.

Now point the same instrument at a person. Why do you need the export? Because I reconcile the numbers. Why do you reconcile them? Because finance asks for it. Why does finance ask? Because the auditor flagged something in March. Why did the auditor flag it?

You are two questions past useful and your colleague has started answering carefully. There was never a single chain here. There were four or five reasons braided together, and by picking one and following it you have quietly discarded the others while feeling rigorous.

This is not a folk objection. In a peer-reviewed critique in BMJ Quality & Safety, Alan Card argues the problem is structural rather than a matter of technique: the method “so grossly oversimplifies the process of problem exploration that it should not be used at all.” His conclusion for root-cause analysis in patient safety is that the five whys “should be abandoned”, and he attaches the aphorism the technique earns: for every complex problem there is an answer that is clear, simple, and wrong.

His scope is healthcare, not software, and I am not going to pretend otherwise. But the defect he identifies is not medical. It is that a chain of five questions produces exactly one causal pathway and then treats the fifth link as the place to act.

The count was never the point anyway

Here is the part that should settle it. The Lean Enterprise Institute, describing Ohno’s own practice, says he asked “sometimes three, sometimes eight, the point being to follow the causal chain all the way down to a root cause.”

So five was never a rule. Which means I cannot honestly tell you that three is the correct number either, and I am not going to. Nothing supports three over five as a matter of causality.

What three is good for is entirely different.

The move

Treat the count as a limit on the person, not on the causality. Three questions is roughly where curiosity starts to feel like cross-examination.

That is a craft claim about conversations, not a finding, and you should hold it as one. But it is the constraint that actually binds when you are asking a colleague rather than a pump, and it points at the right fix: when three whys have not got you there, the answer is not a fourth why. It is a different question.

FIVE WHYS, AT A PERSON            THREE, THEN SOMETHING ELSE

"Why do you need the export?"     "Why do you need the export?"
  to reconcile the numbers          to reconcile the numbers

"Why do you reconcile them?"      "Why do you reconcile them?"
  finance asks for it               finance asks for it

"Why does finance ask?"           "Why does finance ask?"
  the auditor flagged March         the auditor flagged March

"Why did the auditor flag it?"    "What would you have to show
  ...I'd have to check              them to close it out?"

"Why wasn't it caught earlier?"     the before-and-after values,
  ...is this going somewhere?        with dates. which is not an
                                     export. it is a history view
one pathway, chosen at random,
followed until somebody           three whys located the branch.
gets uncomfortable                the fourth question crossed it.

The fourth question is not why. It is what would have to be true, or what did you do instead, or show me the last one. Each of those moves sideways onto a different branch instead of drilling further down the one you happen to be standing on.

What it costs

Three whys will sometimes stop you short of the real thing, and there is no rule that tells you when. You are trading completeness for a working relationship, deliberately. Sometimes that trade is wrong and you will build the wrong thing because you stopped being curious too early.

And the social limit is not universal. In an incident review with engineers, where everyone understands that the object under examination is a system rather than a person, you can go much further and should. The limit is about interrogating people about their own decisions, which is a specific and common situation, not a general law.

Try this week

Take your next why conversation and count. Most people are surprised how far they were going.

Then write your fourth question before the conversation, so you are not improvising it at the moment things get tense. One of these three covers nearly everything:

What would you have to show them to close it out? What do you do instead, today? Can you show me the last one that went wrong?

Ask three whys, then that. You will get to something you can build, and the person will still take your call next week, which is the resource this whole part of the book is actually spending.