Pareto Chart: Build One With a Lab Example

A Pareto chart is a bar graph that displays frequency data from highest to lowest rank order, with a cumulative-percentage line drawn over the bars. It is used to separate the few causes that produce most of the problems from the many that produce little. In a laboratory, the usual input is a tally of event reports, rejected specimens or nonconformances for a set period.

This page gives a complete lookup table from a published laboratory example, the formula and rounding rule behind every column, a second worked example that shows how re-weighting changes the order, and short Excel, R and Python snippets.

What a Pareto chart shows

The AABB Quality Toolkit describes a Pareto chart as a type of bar graph that displays frequency data from highest to lowest rank order. The AHRQ workflow assessment toolkit adds that bar length can represent time, frequency or amount, and that the chart is also called a Pareto diagram. The cumulative line shows where the majority of causes occur.

The idea rests on the Pareto principle, which AABB states as: 80% of consequences come from 20% of causes, the “vital few”. Treat that as a guide to where to look first, not a law. The example below shows why.

Pareto analysis step by step

Pareto analysis is the process of building the chart and reading it. The AHRQ toolkit lists these steps:

List the issues or causes you want to compare. AABB notes that this list typically comes from a brainstorming session, a check sheet or a fishbone diagram (see fishbone diagrams in healthcare).
Choose one standard measure: frequency, cost or time.
Define the data collection period.
Tally each category and sum the total.
Sort the categories from largest to smallest.
Draw the cumulative-percentage line. Its first point lines up with the top of the first bar.
Identify the items causing most of the problems, which AHRQ describes as the place where the cumulative line levels off.

AABB says the chart is helpful when resources are limited, and that it can also be used after an improvement to monitor whether a corrective action worked. For the corrective-action side, see CAPA (corrective and preventive action).

Worked example 1: specimen event reports

The AABB toolkit includes a worked example of specimen event reports totalling 80 events. The table below reproduces its counts and adds the arithmetic for each column.

Rank	Category	Count	Cumulative count	Cumulative %
1	Unlabeled	20	20	25%
2	Hemolyzed	17	37	46%
3	QNS (quantity not sufficient)	14	51	64%
4	Clotted	11	62	78%
5	Phlebotomist info	10	72	90%
6	Two labels attached	4	76	95%
7	Lipemic	3	79	99%
8	Wrong blood in tube (WBIT)	1	80	100%

Source: AABB Quality Toolkit, “Pareto Chart” (checked 2026-10-29). Cumulative % is the cumulative count divided by the total of 80, rounded to the nearest whole percent (for example 37 / 80 = 46.25%, shown as 46%; 51 / 80 = 63.75%, shown as 64%).

How to read the table

The 80% mark falls between rank 4 (Clotted, 78%) and rank 5 (Phlebotomist info, 90%). So the “vital few” here are four or five of eight categories, not literally 20% of them. Four categories (Unlabeled, Hemolyzed, QNS, Clotted) account for 62 of 80 events, which is 77.5%. Real data rarely lands exactly on 80/20, and the rule is a prompt to look at the curve, not a cut-off to enforce.

Formula and rounding

For categories sorted from largest to smallest, with counts n₁ ≥ n₂ ≥ … ≥ n_k and total N = n₁ + … + n_k:

Category share: n_i / N × 100
Cumulative count: C_i = n₁ + … + n_i
Cumulative %: C_i / N × 100, rounded to a whole percent

Round only the displayed value, never the running total. The last row must show exactly 100%. When two categories have the same value, either order is acceptable; the cumulative line will differ only at that one step.

Worked example 2: the same events, re-weighted

AABB warns that a Pareto chart can be misleading if cost or other weighting factors are not incorporated. Its own example makes the point: the source notes that although there was only one wrong-blood-in-tube incident, it should be addressed immediately because of the level of risk. A raw count ranks that event last.

The recomputation below is illustrative only. The weights are made up for this page to show the arithmetic, and they are not data from AABB or any laboratory. Each count is multiplied by a hypothetical severity weight (Unlabeled 3, Hemolyzed 2, QNS 1, Clotted 2, Phlebotomist info 1, Two labels attached 3, Lipemic 1, WBIT 10), then the categories are re-sorted by the weighted score.

Rank	Category	Count × weight	Weighted score	Cumulative score	Cumulative %
1	Unlabeled	20 × 3	60	60	36%
2	Hemolyzed	17 × 2	34	94	57%
3	Clotted	11 × 2	22	116	70%
4	QNS	14 × 1	14	130	79%
5	Two labels attached	4 × 3	12	142	86%
6	Phlebotomist info	10 × 1	10	152	92%
7	Wrong blood in tube	1 × 10	10	162	98%
8	Lipemic	3 × 1	3	165	100%

The total weighted score is 165. QNS drops from rank 3 to rank 4, and wrong blood in tube moves from last to seventh with a score equal to Phlebotomist info. The point is that the order depends on the measure you choose. This is why the AHRQ steps ask you to pick a standard measure (frequency, cost or time) before tallying, and why a low-count, high-risk event should be reviewed on its own whatever its position on the chart.

How to compute a Pareto table in Excel, R or Python

Excel

Put the counts in B2:B9, sorted largest to smallest. In C2 enter =SUM($B$2:B2) and fill down. In D2 enter =ROUND(C2/SUM($B$2:$B$9)*100,0) and fill down. Select the category, count and cumulative % columns and insert a combo chart with the cumulative column on a secondary axis as a line. For the chart-building side, see how to create histograms in Excel.

R

counts <- c(Unlabeled=20, Hemolyzed=17, QNS=14, Clotted=11,
            PhlebInfo=10, TwoLabels=4, Lipemic=3, WBIT=1)
counts <- sort(counts, decreasing = TRUE)
cum_pct <- round(cumsum(counts) / sum(counts) * 100)
data.frame(count = counts, cumulative = cumsum(counts), cum_pct = cum_pct)

Python

counts = {"Unlabeled": 20, "Hemolyzed": 17, "QNS": 14, "Clotted": 11,
          "PhlebInfo": 10, "TwoLabels": 4, "Lipemic": 3, "WBIT": 1}
total = sum(counts.values())
running = 0
for name, n in sorted(counts.items(), key=lambda kv: -kv[1]):
    running += n
    print(name, n, running, round(running / total * 100))

All three produce the cumulative column in the first table. Weighted scores use the same code with each count multiplied by its weight before sorting.

Limits of a Pareto chart

Frequency is not risk. AABB cautions that the chart can mislead if cost or other weighting factors are left out. Example 2 shows how.
Work from the right side. AABB also warns that the chart can mislead if work is done from the wrong side of it, that is, starting with the small bars rather than the vital few.
It ranks; it does not explain. The chart tells you where problems cluster, not why. Pair it with root cause methods such as those in what is root cause analysis, and feed accepted findings into your quality system (see risk-based quality management).
Categories matter. Because the chart is only as good as its tally, define categories and the collection period before counting.

Frequently asked questions

What is a Pareto chart used for?

AHRQ says to use it to find the most significant causes or problems from a larger known set of issues, and to examine how often issues occur in a process. AABB adds that it helps when resources are limited and a team wants to prioritise the causes with the biggest effect.

What is the 80/20 rule in a Pareto chart?

AABB states the Pareto principle as 80% of consequences coming from 20% of causes, the “vital few”. In the specimen table above, the 80% mark falls between rank 4 and rank 5, so the split is approximate.

What is the difference between a Pareto chart and Pareto analysis?

The chart is the bar graph with the cumulative line. Pareto analysis is the process of listing causes, measuring them, building the chart and reading where the cumulative line levels off.

How do you calculate the cumulative percentage?

Divide the running total of counts by the grand total and multiply by 100. For the fourth row of the AABB table, 62 / 80 = 77.5%, shown as 78% when rounded to a whole percent.