Pareto Analysis: Which Cause Should You Fix First?

The fishbone diagram surfaces the possible causes of a problem. But if you leave the meeting with fifteen causes, the real question hasn’t been answered yet: which one do we start with?

Pareto analysis answers exactly that question. It lets the data speak, not opinions.

The Pareto principle

It comes from an observation by the Italian economist Vilfredo Pareto: most of the results come from a small share of the causes. In quality management, Joseph Juran popularised this as the distinction between “the vital few and the trivial many”.

The 80/20 ratio here is not a law of nature but a tendency. What matters is not the number but the idea: don’t spread your effort evenly everywhere.

What is a Pareto chart?

It combines two charts: bars showing the causes in descending order of frequency, and a line showing their cumulative percentage. At a glance you can answer: “if I fix the first few causes, what percentage of the problem have I solved?”

  Count
  150 │ ███
      │ ███                                        ●─── 100%
  100 │ ███  ███                          ●───────
      │ ███  ███                  ●──────
   50 │ ███  ███  ███      ●──────  ███
      │ ███  ███  ███  ███         ███
    0 └──────────────────────────────────────────
       Cust.  VAT  Price  DelN.  Other
        38%   64%   79%    88%   100%   ← cumulative

Example: invoice errors

Say 374 invoice errors were recorded over a three-month period:

The table says one thing: the first three types account for more than three quarters of the errors. Instead of dealing with the remaining seven types, focusing on the customer/supplier account opening process eliminates far more errors with the same effort.

How we build it

1. We clarify the categories. Categories must not overlap. A catch-all box such as “user error” kills the analysis from the start.

2. We choose the right unit. Count is not always the right measure. A frequent but cheap error and a rare but expensive one shouldn’t sit side by side on the same list. Often cost or time lost is the better unit.

3. We take a meaningful period. One bad week can look like a permanent problem. For seasonal work, you need to look at least one full cycle.

4. We sort and calculate the cumulative percentage. We pick the cut-off point by looking at the data; 80% is not a sacred threshold. Sometimes the first two categories make up 65% of the problem, and stopping there is right.

5. We drill down to the root cause. Pareto tells you which category to look at; it doesn’t tell you why that category arises. At that point the fishbone and the five whys come back in.

Common mistakes

The wrong unit. An error that looks harmless by count can jump to the top of the list when you look at cost.

Categories that are too broad. If an “other” box that everything goes into becomes the tallest bar, the analysis is telling you to redefine the categories.

The noise of one-off events. A single major outage can distort the table; we flag such cases separately.

Treating the result as permanent. Once you fix the first cause, the ranking changes. Pareto is not a one-off report but a repeated measurement.

Its place in root cause analysis

Pareto is not the only root cause analysis tool. Approaches such as the five whys, FMEA, fault tree analysis, change analysis and DMAIC belong to the same family; each answers a different question well.

The order we prefer is:

Fishbone  →  Pareto   →  Five whys   →  Action
(causes)    (priority)  (root cause)   (measure)

We first surface the possibilities, then weight them with data, and then drill down to the root of the few causes that remain.

Collecting the data and keeping the chart up to date is a separate job; we usually build it on ErpwareBI and run the analysis work as part of Process Consulting.