Episode 3: Pareto Analysis, Finding the Vital Few

Theo Ashworth

Episode 3: Pareto Analysis, Finding the Vital Few. Imagine a team reviewing reports of accidents, incidents and near misses. They sort the reports into categories and find that a few categories account for much of what was reported. Where should they start investigating?

Nell Harding

The largest categories might be a useful place to start, but a count alone won't tell us which events had the greatest potential for harm.

Theo Ashworth

Exactly. Pareto analysis helps us see how reports are distributed across categories, not decide that the smaller ones are safe to ignore. If the fishbone from last time was about listing possible causes, this is about helping us choose where to investigate.

Nell Harding

Where does the name come from? Because I've always assumed it was a management slogan.

Theo Ashworth

It's an economist. Vilfredo Pareto, Italian, eighteen forty eight to nineteen twenty three. He noticed that roughly eighty percent of Italy's land and wealth was owned by twenty percent of the population. There's also the legend that twenty percent of the peapods in his garden gave him eighty percent of his peas. I'd call that one a legend, because it's the kind of tidy story that gets better with retelling.

Nell Harding

Peas, I like. It makes it sound less like a theory and more like someone looking at his own garden.

Theo Ashworth

The leap to quality came from Dr. Joseph Juran. In 1941 he recognised that the same lopsided pattern showed up in defects, in engineering, in management. He called it the vital few and the trivial many. Later he changed that to the vital few and the useful many, so managers wouldn't hear the word trivial and decide the small contributors were worthless.

Nell Harding

Now, I'll push on the numbers. Is it always eighty twenty?

Theo Ashworth

No. The two figures measure different things. One is the share of categories, the other is their share of the reports. The distribution might be lopsided, or it might not be. It's a rule of thumb, not a law of nature. Your data decides, not the slogan.

Nell Harding

Fine. So how do I actually build one?

Theo Ashworth

Start with categories and counts from your accident, incident or near miss reports. Be clear about what each category means and what you are counting. Then draw a bar chart with the categories along the bottom, sorted from biggest on the far left down to smallest on the right. Lots of tiny categories can be batched into one bar called Other, placed at the far right, but don't let that hide a serious risk. The left vertical axis shows the count.

Nell Harding

And the line?

Theo Ashworth

The right vertical axis runs from zero to a hundred percent, and you plot the running total. First dot sits at the top of the first bar. Then you add the second category's share, then the third, and keep going until you reach a hundred percent at the last bar. For a simple illustration, if four categories represent twenty nine, twenty five, twenty one and eleven percent of reports, the running total reaches fifty four percent after two and eighty six percent after four.

Nell Harding

So reading it is about the shape of that line.

Theo Ashworth

Yes. A steep climb at the start means a few categories account for a large share of the reports. Where the curve starts to flatten can help you identify a manageable set to investigate first, while keeping serious risks elsewhere in view.

Nell Harding

Here's what bothers me. There's not always a clean bend, is there? What if the curve just slopes gently?

Theo Ashworth

The Juran guide calls that the awkward zone, where there's no clear breakpoint. The cumulative line helps you see how the categories add up, but it doesn't always give you an obvious place to stop. You'll need judgement as well as the chart.

Nell Harding

Then there's the trap I think catches people out most. What are you counting?

Theo Ashworth

That's the big one. Say you chart workplace injuries by count. Minor scrapes on hands and arms might happen hundreds of times, with negligible cost and danger. A serious head injury might happen once. On a frequency chart, the scrapes tower over everything and the head injury is a sliver on the right, hidden in Other.

Nell Harding

And that serious injury still deserves attention, even if it barely registers on the frequency chart.

Theo Ashworth

The same goes for a near miss with the potential for serious harm. You can compare a chart by frequency with another view based on severity or potential severity. That can bring different priorities into view. A rare, serious event shouldn't wait just because its bar is small.

Nell Harding

So Pareto answers where. What it doesn't answer is why.

Theo Ashworth

Exactly. A Pareto chart won't tell you the actual cause, only where to start digging. And it can mislead you if you lump categories together the wrong way. A broad category of incidents might hide very different events and circumstances.

Nell Harding

How do you handle a bar that broad?

Theo Ashworth

You drill down. It's called a nested Pareto. Take the reports in that broad category and chart their subcategories. Now you've got a more specific picture of where to investigate.

Nell Harding

And that's where the earlier episodes slot in.

Theo Ashworth

They do. Pareto can help you choose incident categories to investigate. Then you can use an Ishikawa diagram to explore how those events might happen. The 5 Whys can help the team explore a plausible branch further, checking it against evidence. Each tool covers a different part of the investigation. The Pareto doesn't establish the cause on its own.

Nell Harding

One more use, because I think it gets forgotten. Checking what happened after a corrective action.

Theo Ashworth

Yes. Make the chart again after the corrective action and compare the counts over comparable periods. A change can prompt useful questions, but the chart alone doesn't prove the action caused it. Reporting practices can change too, especially for near misses.

Nell Harding

So the habit is simple. Count first, rank it, and only then start asking why.

Theo Ashworth

Count first. Before someone's favourite theory gets a task force.