Why – why, why, why, why?

Five whys is a method of finding the root cause of a problem. It was developed within Toyota – either by Sakichi Toyoda or Taiichi Ohno, depending on the website you find – and continues to be one of the root tools used by that company to find root causes of issues and problems. There are lots of different approaches to this simple tool for finding a root cause for an issue or problem – branched Five Whys, Multiple Start, “3 Legged”, etc. All of these are great expansions of the basic concept – and it is the basic concept that we need to study first.

Why 5?

The first question is, why five? What if we find the root cause on the first “why”? Or the third? What if the “Whys” start becoming circular, pointing to themselves? Who decided 5 is the right number? Maybe my company only needs 3 whys.

The real answer lies in the tried history of the method. It is true, there will be times when you really find a root cause in under 5 iterations – and there will be some where 5 isn’t enough. Nobody will fault you for asking “Why” for a sixth time – but an auditor will absolutely fault you for stopping at 4 unless you have a clear reason and you have a defined process that allows you to do so.

Author’s note: this is one of the issues with auditors, BTW. They will ask for things like 5 whys when there might not be a fifth, or they will ask for a measurement for a part that needs no measurement. What they really want is something in your process that documents how exceptions are handled. But I digress…

In actual use and repeated practice the good folks at Toyota found that asking 5 times will drive deep enough into a problem to truly root out what is important. When you use the technique you should do your level best to look five levels deep into the problem.

The problem lies in a process

The answer to a Five Whys question cannot be an individual. In other words, if a machine was delivered to a customer and the gears were not treated with the first and essential coating of lubricating grease, the question could be “Why were the gears not greased?” The answer cannot be “Because Ed didn’t grease them.” This may be a true statement – Ed did not, in fact, grease the gears. But unless Ed came to work that day intending to do harm to the company, the fault is not with Ed. The fault is with the system and the process that allowed these questions to remain unanswered:

• Why did Ed not know to grease the gears?
• How were the gears checked to make sure they were greased?
• Why did the rest of the gear train get assembled with ungreased gears?
• Why did nobody else notice the gears were not greased?

As you see, there are lots of questions that could have replaced “Why were the gears not greased?” that would have given a different answer than “Because Ed didn’t grease them.” The problem isn’t Ed. The problem is all around Ed, but Ed is not the issue needing a solution.

I know – loads of folks reading this have situations where they replaced one individual who was trained and experienced and competent with someone else and the gears didn’t get greased (or whatever your particular failure mode might be). It remains true, though, that the fault is not with the person. The proof of this is because you replaced the person and things went wrong – that meant your system isn’t robust enough to ensure good quality results. Instead of making sure the process is bulletproof, you have relied on an individual and therefore put the company at risk. This is the essential reason the fault cannot be with the individual. In fact – when you have a process that depends on the skill and experience of an individual you have a process with a built-in failure point. Processes that are built around an individual are number one targets for improvement to build that individual out so they can take a vacation, but build that person’s techniques in.

Each question should be studied

It is not true that any question asking “Why” is a fair question. As demonstrated in the previous case, the question “Why were the gears not greased?” might not be the first question. If the first question is “Why did the gears leave assembly without grease on them?” the answer changes a lot. This answer might be “There was no grease in the kit”, or “the drawing doesn’t say to grease the gears”, or “there is no process for greasing gears”. Maybe Ed just knows to grease the gears. That means if Ed is out sick the person filling in won’t know to do this. Hey – maybe Ed was out that day! Now the answer is “Ed was out that day, and Bob didn’t know to grease the gears.” See how important that first question can be? Take time with the question.

Sometimes the Five Whys trail leads in a strange direction that cannot be addressed procedurally. That’s an indicator that a wrong question was introduced, and the Five Whys trail might need to back up. Start walking back through the questions and study each one – was the question phrased such that it looked at the process – and was not reliant on an individual?

1. Why were the gears not greased?
   • The drawing does not specify when and how the gears should be greased, it just calls out the grease. There is no instruction to apply grease.
2. Why is there no instruction to apply grease?
   • We typically rely on Ed to apply the grease, he is experienced and knows to do this.
3. Why do we rely on Ed’s experience instead of documenting the process?
   • It has never been a problem before – Ed has been doing this for 14 years.
4. Why do we rely on Ed?
   • It has never been a problem before – Ed has been doing this for 14 years.
5. Why do we rely on Ed?
   • It has never been a problem before – Ed has been doing this for 14 years.

Notice at the 3rd question we fell into a trap. We allowed Ed to be the fault. An individual is never the fault, so let’s look at the answer to number 2. “We rely on Ed.” This is a true statement, but it is not the answer to the question. The question was “Why is there no instruction to apply the grease?” The answer to that question is more systemic. Here are some possibilities:

• We don’t use work instructions
• We have never followed this process to see what instructions should be created
• This is the first time we ever built this piece of equipment, the instructions were not created before release

Oh – that last one is a hot switch. The first “We don’t use work instructions” is a bad answer in general – it means you have practically zero ability to train a new employee. Let’s consider option three and see where it leads…

1. Why were the gears not greased?
   • The drawing does not specify when and how the gears should be greased, it just calls out the grease. There is no instruction to apply grease.
2. Why is there no instruction to apply grease?
   • This is the first time we ever built this piece of equipment, the instructions were not created before release.
3. Why were the drawings released without work instructions?
   • We only develop work instructions for mature processes.
4. Why are work instructions only made for mature processes?
   • We don’t know what work instructions are needed until we have run the process a few times.
5. Why do we not know what processes are needed before we release to production?
   • We don’t do a “prototype” or a mock up of our processes, and we don’t carefully monitor our processes during the first few runs.

This is a candidate for another “Why” – “Why do we make production equipment with untested processes?”, but this last and fifth answer points straight at the cause. This particular failure came about because the company culture is to build first and ask questions later. If the process was carefully watched with an eye for possible failure and places where work instructions will be needed, the end result would have been a well-greased set of gears.

Many causes, one root

There are opinions out there that a Root Cause Analysis will always have more than one root cause. This could be true, but more likely the reality is that there are many contributors to a condition – but one of them is either the major contributor or the one thing that would override all the other contributors. Look at the example – Grease was not applied to the gears. The root cause, the thing that ultimately created the conditions for this to happen, is this: when the company creates a new process to make a new thing the process is not studied to make it “bulletproof”. Instead, assumptions are made that folks will “do the right thing”. But along the way some other observations came to light that are contributors, certainly, but not the root cause.

1. Drawings are released without work instructions. If the culture was based around having work instructions for critical steps, and maybe a risk analysis to tell folks what steps are critical, this might have been avoided.
2. The culture makes assumptions about mature processes. This is a breeding ground for reliance on individuals over process since an individual might be really good at something and nobody pays attention to how this might affect the output until that individual is absent.
3. Grease is called out with no instructions regarding where and how to apply the grease.

So as the analysis proceeded there were a hatful of improvements that turned up to address contributing factors, but there was one root cause.

The basics of a Five Whys analysis are important – more important than all the variants of the method. You need to know how to ask the questions, as much as you need to know the answers. A problem well defined is half solved. Keep a sharp eye on the process, check the questions, and never allow the answer to be an individual.