Pareto Chart Maker — Counts, Cumulative Percentage and the 80% Line
A Pareto chart maker that runs online in your browser: paste a frequency table or a raw occurrence log, and it sorts the categories, tallies them if they need tallying, draws the cumulative-percentage line on a second axis and marks the 80/20 cut.
What the chart is actually telling you
A Pareto chart answers one question: of everything going wrong, which few things account for most of it? The bars are counts by category, tallest first. The line climbing across them is the running total as a percentage, so wherever it crosses your reference level — 80% by convention — you can read off how many categories you would have to fix to get there.
That number is the output. Not the tallest bar, not the shape of the curve: the count of categories inside the line, and the percentage they actually come to. This tool prints both under the figure, because "the first three of nine causes are 85% of the complaints" is a sentence you can take to a meeting and "here is an 80/20 chart" is not.
The 80/20 framing is a rule of thumb and this tool treats it as one. The reference line is a setting, the figure reports the cumulative total genuinely reached rather than the number you asked for, and when it takes six of ten bars to get there the chart says so in as many words — because a flat distribution has no vital few, and ranking it will not help you.
Data
Paste straight from Excel, Minitab or a CMM export — include the header row and the columns are matched by name, in any order.
| # | Category | Count | |
|---|---|---|---|
| 1 | |||
| 2 | |||
| 3 | |||
| 4 |
0 rows of data
Loading the calculation engine…
Enter data on the left to chart it
SVG is vector — it stays sharp at any size in a report. PNG 4× is roughly 600 dpi at figure width. Every export carries the estimator, the constants and the rule set drawn inside the figure, so the file is readable on its own.
Downloads carry a small “made with spcplot.com” credit along the bottom edge. The chart on screen is never watermarked, and the credit is a credit rather than a licence restriction — the figures are yours to publish either way. A Pro or Lifetime licence removes it.
Options
This plot has no adjustable options yet.
Your measurements are processed entirely in this browser tab. Nothing you type here is uploaded, so confidential production data stays confidential.
Two ways to give it your data
A frequency table. One row per category, with the count beside it. This is what you have if somebody has already summarised the month.
A raw occurrence list. One row per event, category only, Count column left empty — a complaint log, a scrap ticket export, a downtime reason column copied straight out of the MES. The rows are tallied for you. Retyping four hundred lines into a pivot table before you can chart them is the chore this removes.
Left on Automatic the tool takes the first reading if anything numeric appears in the Count column and the second if nothing does, and it prints which one it used under the chart. The two readings of the same paste differ by orders of magnitude, so that line is not decoration. Repeated categories are added together either way, and a row whose count is blank or unreadable is dropped and counted rather than quietly read as zero or as one.
The “other” bucket, which is where a Pareto chart can mislead
A long tail of one-off categories crushes the bars that matter, and merging the tail into one bar is the standard fix. It is also the only real editorial decision on this chart, because whoever sets the threshold decides what the reader is allowed to see. So the rule and everything it absorbed are printed under the figure: how many categories were merged, what they came to, and by which rule.
Three rules are offered — keep the top N, merge anything below a share of the total, or merge everything past a cumulative percentage. A tail of a single category is never merged, because that would only delete its name from the chart.
The merged bar is always drawn last, whatever its height. If "Other" comes to more than the fourth-ranked cause it still goes at the right-hand end, because the descending order is an order of *categories* and "Other" is not one — putting it in rank position kinks the cumulative line and invites somebody to go and fix a bucket that cannot be fixed. The size of that bar is itself a finding: if it is large, your categories are too fine and the data-collection form is the thing to change first.
Reading the figure
Bars are counts against the left axis; the black line with open markers is the cumulative percentage against the right axis, plotted above the centre of each bar. The dashed horizontal rule is the reference level and the short dashed vertical is the divider after the last category needed to reach it. Bars inside that divider are drawn at full strength and the rest are faded, which changes no number and makes the recommendation visible from across a room.
By default the count axis is scaled so the grand total sits at the top, which puts 100% and the total at the same height and makes the line genuinely comparable with the bars. Switching to the independent scaling stretches the count axis to the tallest bar so that small categories stay readable — and the figure then states that the two axes are no longer comparable, because that is exactly the sort of thing a reader will otherwise assume.
What this tool does not claim
Nothing here is estimated. There is no control-chart constant, no sigma, no fitted parameter and no distributional assumption anywhere in the calculation — it is a sort, a running total and a division. That is why the validation page carries no reproduced worked example for this chart and says so plainly: there is no published answer to reproduce, so the test suite pins the arithmetic to a dataset of five hundred complaints whose every percentage is a whole number and which a reader can check by hand in a minute.
A footnote worth knowing, since the name is on the tin: Vilfredo Pareto studied the distribution of wealth and never drew this chart. J. M. Juran attached Pareto’s name to the "vital few and trivial many" and later published a correction — "The Non-Pareto Principle; Mea Culpa", in *Quality Progress* — saying the generalisation had been his own and the attribution was a mistake. It stuck anyway.
Common questions
- Can I make a Pareto chart from a raw list without counting anything first?
- Yes. Paste the category column on its own and leave Count empty — one row per occurrence — and the rows are tallied. The figure says it tallied rather than read counts, so nobody has to guess which happened.
- Why does my 80% line land at 85%?
- Because the cumulative total jumps by whole categories. The tool includes the first bar that reaches or passes your reference level and then reports the cumulative percentage actually reached there, which is almost never the reference itself. Reporting “80%” when the truth is 85% would be rounding the data to fit the slogan.
- How many categories should I show before merging the rest?
- There is no correct answer, which is why the threshold is yours and the merge is disclosed on the figure. As a working rule, if the merged bar is bigger than your third-ranked cause then the categories are too fine and the collection form needs fixing more than the chart does.
- Can I make a Pareto chart of cost or downtime rather than counts?
- Yes, and it is often the more useful chart — a hundred cheap defects can matter less than three expensive ones. Put the cost or the hours in the Count column. Decimals are kept and the axis label is yours to change.
- Does this need a control chart first, the way capability analysis does?
- No. A Pareto chart makes no assumption about stability, normality or anything else; it is a ranked summary of what you counted. It is usually the first thing you draw, and the control chart comes afterwards on whatever you chose to work on.
- Is my data uploaded anywhere?
- No. The whole calculation runs as JavaScript in this browser tab, and nothing on this site accepts a measurement — the only server code here takes a payment. That is why the tool is usable on data you are contractually barred from putting in a cloud service.