Some thoughts on “5 Whys”

Jason Yip
4 min readJun 26, 2022

--

The “5” is a reminder to avoid shallow attribution

When faced with a problem, there’s a general tendency for people to only consider the immediate contributing factors.

Service stopped working because SSL certificate expired. “Well, we can’t stop SSL certificates from expiring, I guess there’s nothing we can do!”
People have a general tendency for shallow attribution

The “5” in “5 Whys” is a reminder to go beyond shallow attribution and keep exploring underlying causes.

Service stopped working. Why? We didn’t renew the SSL certificate before it expired. Why? We didn’t know that the SSL certificate was about to expire. Why? …
5 Whys is a reminder to go beyond shallow attribution

“5” is a heuristic not a fixed number

In practice, the number of whys can be less or more than 5. Because of the tendency to be shallow, I’d still suggest thinking about it more if it ends up being less than 5.

“5” is not a fixed number but more a heuristic

5 Whys, not 5 Whos

The use of “why” is to remind us not to default to blaming people, especially their motivation (aka fundamental attribution error), and consider broader factors for why something happened.

“Who” focuses only on personal vs other factors

See also Six Sources of Influence.

“How” seems clearer than “Why”

The purpose of root cause analysis is to understand mechanism(s) to allow us to come up with better ideas for how to intervene. “How” expresses this more clearly than “Why”.

“Why” can be confused for asking about motivation; “How” only implies mechanism
“Why” can be confused for asking about motivation

See also “Why” is about purpose; “How” is about mechanism.

But there’s usually more than one root cause!

There is no assumption that there is a single root cause. It’s not called “single root cause analysis”.

The idea is that you’re exploring a single promising factor with 5 Whys. This does not mean you won’t explore more.

“Should we also explore why it took so long to detect the problem?” “Nah, I’m pretty sure we’re not supposed to think for ourselves with 5 Whys.”
It’s not “single root cause analysis”

To reliably interrupt a mechanism, intervene at multiple points, not just the “root cause”

If you want to reliably interrupt a mechanism, you should do it at multiple points. This is because no countermeasure is perfect and layered defense is an effective meta-countermeasure for this.

Diagram showing interventions at multiple points in a 5 Whys chain
Intervene at multiple points

Example borrowed from The Five Whys for Start-Ups by Eric Ries.

You don’t actually need to address every contributing factor

You don’t actually need, nor want, to address every contributing factor because the possible countermeasures may be expensive, and unnecessary if other cheaper factors are addressed.

We’re seeing more random blow-ups over small things. Explore factors we have influence and leverage. Don’t explore factors where we have limited influence.
You don’t actually need (or want) to explore every contributing factor

But it’s too simple to work for complex problems!

Every organisation has problems at different levels

If you look at any organisation, problems are happening all the time at different levels. Typically, in terms of numbers, most problems are simple, the least problems are complex, and complicated problems are somewhere in the middle.

The least number are sophisticated, complex problems. Somewhere in the middle are complicated problems. The largest number are day-to-day, simple problems.
Distribution of problems in most organisations

Use the simplest method that is useful (including fitting within the time you have available)

Simple problems can be addressed with simpler methods (like 5 Whys), while complicated and complex problems warrant more sophisticated methods. It’s not useful to apply the same root cause analysis / problem-solving approaches for complex problems as you would simple problems.

“5 Whys most applicable here” pointing to the day-to-day, simple problems
5 Whys is most applicable to day-to-day simple problems

5 Whys is not intended for complex problems; sophisticated, time-intensive causal analysis is not intended for simple problems.

Service stopped working. Why? SSL certificate expired. “Wait! Before we go further, let me set up an anthropological study…”
More sophisticated approaches are not intended for simple problems

To learn more

Find some 10+ year Toyota person to talk to. They’ll probably have more insight than I have.

--

--

Jason Yip

Senior Manager Product Engineering at Grainger. Extreme Programming, Agile, Lean guy. Ex-Spotify, ex-ThoughtWorks, ex-CruiseControl