Problem solving Explore

Fishbone Diagram and 5 Whys

The fishbone diagram and 5 whys lay out the possible causes of a problem by category, then keep asking why to reach the cause worth fixing.

The questionWhat categories of cause could produce this effect, and which cause, if fixed, would stop it happening again?

The fishbone is also called an Ishikawa diagram or a cause-and-effect diagram. The 4M version uses man, machine, material, and method. Five whys is also written 5 Whys.

Process · Kaoru Ishikawa and Taiichi Ohno, 1960s and 1970s

ManSkills, staffing, and attention
MachineEquipment and tools
MaterialParts and inputs
MethodProcedure and sequence
MeasurementGauges and data
EnvironmentLayout, climate, and demand

The effect sits in the center. Each spoke is a category of possible cause. Ask why along a spoke until the cause is one you can change.

When to use it

A defect, delay, or failure has an obvious first explanation, and that explanation has already been "fixed" without the problem stopping.
A group is brainstorming causes with no structure, so the list is long, overlapping, and still missing a whole category such as measurement.
You need a shared picture of possible causes before anyone collects data, so the next observations have somewhere to land.
A lean, TPM, or A3 review needs a cause analysis that people on the floor already recognize.

When to avoid it

The problem is a defined deviation with is and is-not facts, and you need to test causes against those facts. Kepner-Tregoe problem analysis is tighter.
You already have counts by cause. Rank them with Pareto instead of another brainstorming session.
Cause and effect cannot be traced in a stable chain, as in a Complex situation. Five whys will produce a story that fits the past and fails the next time.
The sponsor wants a single "root cause" for a problem that has several independent contributors. The diagram will hide that if you force one spine.

Overview

A fishbone diagram sorts possible causes of a stated effect into a few categories, so a group can see which kinds of cause it has not yet considered. Five whys then takes one cause and walks it down until the team reaches a condition it can change. Used together, they stop the first explanation, usually an operator or a missing procedure, from becoming the whole analysis.

The effect has to be a fact with a count, and the last why has to be checkable on the floor. Five is a prompt, not a depth. Empty categories are a finding. A chain that ends at human error, culture, or a cause nobody can observe is not finished, and a single spine is the wrong picture when two independent causes each explain part of the gap.

The diagram and the chain

The effect

A specific, observable problem, written as a fact with a count or a date. "Late shipments" is a topic. "Twelve percent of orders missed the promised day last month, against a three percent target" is an effect. Vague effects produce vague bones.

The categories

A small, agreed set of cause types. Manufacturing teams usually use 4M or 6M. Service teams should rename the bones so people will actually use them. Empty categories are a finding. They mean the group has not looked there yet.

Causes and subcauses

Each idea is filed on a bone, then split if it is still a symptom. A cause that could sit on two bones should be written once, with a note, not copied, or the later fix will be duplicated.

Five whys

Pick a cause that looks worth testing and ask why it is true, then why that is true, until the answer is a condition the team can change and that would prevent recurrence. Stop early if you reach a changeable cause. Keep going past five if you are still on a symptom.

What it is not

A proof of the first story the room believed. A fishbone that only has bones under Man, or a five-whys chain that ends at "human error," has not done the work.
A substitute for data. The diagram holds hypotheses. Counts, is-and-is-not facts, or a trial still have to test them.

The steps

Write the effect as a fact

Name what is wrong, where, how much, and over what period. If the group cannot agree on the sentence, you do not yet have a problem, you have a topic.
Set the categories

Choose 4M, 6M, or a service set the people in the room will use. Draw the spine and the bones before generating causes, so ideas have a place to go.
Fill the bones

Ask what in this category could produce the effect. Go around the people who run the process, not only the people who manage it. Mark empty bones and fill them before you stop.
Walk five whys on one or two causes

Pick the causes that, if true, would explain the size of the effect. Ask why until you reach a changeable condition. Write each link as a fact that could be checked on the floor the same day.
Test, then fix the cause you reached

Confirm the last link with a measurement or an is-and-is-not check. Assign the fix, a date, and the measure that would show recurrence. Do not close the problem on the workshop vote.

In practice

A food plant's fill line was rejecting 4.2 percent of packs against a 1.1 percent target. The first workshop blamed operator skill and booked two days of training. The consultant redrew the effect as "4.2 percent seal rejects on line 3, last four weeks," and used 6M. Machine and Measurement were almost empty, which was the clue.

Five whys on Method, with the maintenance lead in the room, ran like this. Why do seals fail? Torque on the sealing jaws is below spec. Why? The torque wrench reads in spec on the line and out of spec in the shop. Why? It has not been calibrated in fourteen months. Why? There is no calibration schedule for tools that stay on the line. Why? Nobody owns gauges that are not in the metrology cage. The plant named a metrology owner, put the line wrenches on a weekly check, and rejects fell to 1.3 percent in six weeks. The training would have left the wrench as it was.

What you leave with

An effect statement with a count and a time window.
A filled fishbone, including the categories that started empty.
One or two why-chains that end in a changeable cause, each with the evidence that supports the last link.
A named fix, owner, and measure for recurrence.

Failure points

Human error as the last why

The chain stops at an operator. Operators work inside a method, a machine, and a measurement system. Stopping there produces retraining and the same defect next month.

Five as a quota

The team asks why five times even after it has a changeable cause, or it stops at five while still on a symptom. Depth follows the chain, not the number.

One bone filled

Everything is filed under Method or Man because those are the words the group already uses. The empty bones are where the missed cause usually sits.

A chain nobody can check

Each why is a story about incentives or culture. Nobody can go to the line and see whether it is true, so the fix is a poster.

Single root cause for a compound problem

Two independent causes each explain part of the effect. Forcing one spine hides the second, and the metric only half-recovers.

The failures above are usually Framing: the wrong problem was chosen before this framework was applied.

Origin

Kaoru Ishikawa, a professor at the University of Tokyo, developed the cause-and-effect diagram in Japan. He is often said to have first used it at the Kawasaki shipyards in the 1940s, and it was in wide use as a quality tool by the 1960s. English readers met it as one of the seven basic tools in his Guide to Quality Control. The usual manufacturing categories are the 6Ms, man (or manpower), machine, material, method, measurement, and environment (mother nature). Service versions substitute categories such as policies, procedures, and people. Nobody owns the diagram, and it carries no trademark.

Five whys was described by Taiichi Ohno as the basis of Toyota's scientific approach. In Toyota Production System he wrote that by repeating why five times, the nature of the problem and its solution become clear. The practice is often traced earlier to Sakichi Toyoda. Five is a rule of thumb, not a required depth. The method is public and unlicensed. Used together, the fishbone collects candidate causes by category, and five whys walks one candidate down to a cause the team can change.