Fishbone diagram

The problem goes in the head of the fish, at the right. A spine runs back from it, and bones branch off the spine, one for each category of cause. Possible causes hang off each bone, and causes of those causes off them in turn. The finished diagram looks like a fish skeleton, which is how it got its name.

It shows possibilities, not proof. Every cause on it is a suspect until data says otherwise. Its value is that the whole team sees the whole list at once, and nobody's favourite theory gets fixed before the others have been considered.

Use it when a problem is clear but its causes are not, and when several people each hold part of the answer: the nurse, the pharmacist, the porter and the manager all see a late discharge differently. It is a working session tool. Drawn on a wall with the people who do the work, it takes about an hour.

It is the natural next step after a Pareto chart. The Pareto chart says which category of problem to work on; the fishbone digs into why that category happens.

Not when the cause is already known and agreed; then the job is to fix it and measure. Not as the end of the analysis: a fishbone full of plausible causes is a list of hypotheses, and the next step is to test the likeliest ones with data. And not when the problem statement is vague. A fish with "poor performance" in its head grows bones in every direction.

Manufacturing traditionally uses six headings known as the 6Ms: machines, methods, materials, measurement, manpower and mother nature, meaning the environment. Service work is better served by headings that fit it: people, process, policy, place, systems and measurement. Use whichever set makes the team think of causes it would otherwise miss, and add or drop a heading when the problem needs it.

Look for the causes that appear on more than one bone, which often point at a single root. Look for bones that are nearly empty, which usually mean nobody in the room knew that part of the process, not that it is free of causes. Then mark the few causes the team thinks most likely, and decide what data would confirm or rule out each one.

The ward from the Pareto chart page found that three causes, medicines to take home, transport and the discharge letter, accounted for most of its late discharges. It drew a fishbone with the problem "discharges finishing after noon" in the head, using the service headings. The causes below are illustrative, of the kind such a session produces.

Two causes appear on several bones: medicines being ordered on the morning of discharge, and nobody owning the discharge on the day. The team chose to test the first, ordering medicines the afternoon before. The run chart page shows what happened next.

Write the problem as a specific, measurable statement in the head. Draw the spine and one bone per heading. Ask why the problem happens, under each heading in turn, and write each answer on its bone; for each answer, ask why again. Stop when the answers become things the team could test or change. Then circle the likeliest causes, and plan how to check each with data before anything is fixed.

The fishbone belongs to Analyse, between the Pareto chart that chose the problem and the data collection that confirms the cause. It often reappears in Improve, when a change does not work and the team returns to its list of suspects.