The 80/20 Principle: A Complete Guide

You export a year of invoices, sort your clients by what each one paid, and add a running total. A few names in, the total has already passed half of the year’s income, and the long tail of small clients at the bottom barely moves it. The top of the list is much shorter than you expected.

That lopsided list is what the 80/20 rule, also called the Pareto principle, describes: a minority of causes produces most of an effect. The pattern turns up again and again in real data. The exact ratio does not, and the productivity version, that a fifth of your tasks yields four fifths of your results, has not been tested directly in any study we could find.

Used as a habit rather than a law, the rule comes down to three moves, each explained below: sort your own numbers largest first, trace what the top few items depend on, and sort again later, because the top few change.

What the 80/20 rule actually claims

The 80/20 rule claims that when many causes add up to one result, a small share of the causes accounts for most of the result. Joseph Juran, the quality engineer who gave it the name Pareto principle, described it as the vital few and the trivial many, and saw it in defects, absenteeism and the causes of car accidents.¹

The numbers are a label, not a measurement: one is a share of the causes, the other a share of the effect, and any pairing is possible. Where the rule has been checked against large data sets, the imbalance is real and the ratio moves. Across 238 US grocery brands tracked from 2004 to 2009, the top fifth of each brand’s buyers brought in about 73 percent of its sales on average, according to a 2017 study by Baek Jung Kim, Vishal Singh and Russell Winer of New York University.² That is most of the money, and it is not four fifths.

The rule is one of several ideas that productivity methods borrow; an overview of the best-known productivity methods covers the methods themselves, from GTD to time blocking. For your own planning, treat 80/20 as a question (how lopsided is this list?) rather than an answer you already know.

Running a Pareto check on your own numbers

A Pareto check sorts causes by size and follows the running total, the method Joseph Juran used on his lists of defects. It shows how few causes carry most of the effect in your own case, so you work from your ratio instead of assuming the classic one.¹

Sorting works because totals hide shape. A monthly count of support tickets says nothing about whether one cause drives half of them or dozens of causes drive a few each. A sorted list with a running total shows which at a glance; where the total stops climbing fast, the top group ends.

1. 1List the causesclients, task types, error types
2. 2Count one outcome for eachmoney, hours, complaints, rework
3. 3Sort, largest firstand watch how fast the running total climbs
4. 4Trace the top fewlist everything each one depends on
5. 5Count again laterthe top few change

Picture a support team lead who wants fewer repeat tickets. She tags a month of tickets by cause and sorts the tags. Two causes, password resets and a confusing invoice layout, turn out to generate most of the volume. Before redesigning the invoice, she lists what the change touches: billing software, the help-center article, the finance team’s templates. Next month she sorts the tags again to see whether the order has changed. The same loop sits at the heart of how process improvement works: measure, change one thing, measure again.

To run steps 1 to 3 on your own list, paste it into the box below.

Steps 4 and 5 are the ones people skip; the software and health-spending evidence below shows why they matter. Juran’s later wording points the same way: a summary in the archived copy of his 1974 essay notes that the trivial many was later changed to the useful many.¹ Dropping the tail is a decision to test, not a free saving.

Where the name came from, and why Juran regretted it

The Pareto principle is named after an economist who studied incomes, not effort. It has three layers, each younger than the one before: Vilfredo Pareto’s income data, Joseph Juran’s general rule, and a productivity writer’s claim about personal effort.

Pareto plotted cumulative income data for England, several Italian cities, German states, Paris and Peru on double logarithmic paper, claimed that each gave a straight line with about the same slope, and asserted a law of income distribution, as the economist Joseph Persky summarized in 1992.³

Juran supplied the generalization. As a young engineer in the mid-1920s he noticed that when a long list of defects was sorted by how often each occurred, a relative few accounted for most of them. Writing the first edition of his Quality Control Handbook in the late 1940s, he needed a short name for a pattern he had come to see as universal, noted that Pareto had found wealth to be unequally distributed, and borrowed his name. When other quality experts challenged the attribution, Juran went back to Pareto’s work and concluded that it belonged to economics.¹

his models were not intended to be applied to other fields

The publisher of Richard Koch’s bestseller The 80/20 Principle describes its premise as most results in business and in life coming from a small share of effort, and quotes the book’s conclusion that four fifths of effort is largely irrelevant.⁴

Myth

Pareto discovered that 20 percent of effort produces 80 percent of results.

Fact

Pareto studied the distribution of income. Joseph Juran generalized the pattern, named it after Pareto, and later said the name was a mistake.

Knowing the layers tells you how much weight each use can bear. Pareto’s income curves were measured, and Juran’s rule rested on years of sorted defect lists, so a claim about customers or faults stands on firmer ground than a claim about your own effort, which is the newest layer and the least tested.

Why a few causes dominate, and why the ratio moves

Lopsided results reflect what statisticians call heavy-tailed distributions, where a handful of very large values sit beside many small ones. How much the top fifth takes depends on how heavy that tail is, so the ratio changes from one data set to the next. The physicist Mark Newman set this out in a 2005 review of power laws.⁵

Newman showed that for a power-law distribution, the share held by the top follows from a single measure of its steepness. With the steepness estimated for wealth, the richest fifth of people should hold about four fifths of the wealth, the classic ratio. The same arithmetic applied to websites gives the busiest fifth about two thirds of all hits. Both are lopsided; only one matches the classic ratio.

Across: causes, largest first, from none to all. Up: share of the total effect. Dashed line: the top fifth of causes.

1. Equal shares: every cause contributes the same, so the top fifth of causes gives a fifth of the effect
2. Gentle skew: the top fifth gives more than its share, but well short of most
3. Steep skew: the top fifth gives most of the effect; this is the shape the 80/20 rule describes

One common source of heavy tails, Newman explains, is rich-get-richer growth: a paper that already has many citations is more likely to be found and cited again, and the same process has been used to explain links between web pages.⁵ Your inbox shows the everyday version. Most messages are routine, one or two carry a decision, and no ratio tells you in advance which ones.

Knowing which curve you are on changes the plan. On a steep curve, a few items deserve close attention; on a gentle one, broad effort still pays. The only way to tell is to sort your own list.

Software bugs: the busiest files are rarely fixed alone

In software, a minority of files draws most bug fixes, but the fixes that touch those files commonly change less frequently fixed files too, so working on the top few alone would miss much of the work. That is the finding of a 2018 analysis of 100 open-source projects by Neil Walkinshaw and Leandro Minku.⁶

Earlier studies, starting with Norman Fenton and Niclas Ohlsson’s in 2000, found faults concentrated in a minority of modules, but each looked at one to three industrial telecoms systems.⁶

The study

Limited evidence

The top fifth of files, and everything their fixes touched

Counting each file edited in a fix as one defect, the most-fixed fifth of files accounted for about 80 percent of fixes. But most fixes changed several files. Counting every file those fixes changed, the files behind the same share of fixes made up about a third of the median project’s files, not a fifth, and a power law did not fit the spread of fixes in about a third of the projects.⁶

The authors conclude that focusing on the most-fixed files would probably be misguided, because the fixes that involve them commonly include less frequently fixed files. The main caveats: it is an observational studyobservational study: A study in which researchers record what people already do or are exposed to, rather than assigning anyone to anything. It can show that two things go together, not that one causes the other, because the groups being compared may differ in other ways as well.Full entry in the glossary, bugs were identified automatically from words in commit messages, and every project was open source.⁶

Take a consultant whose two largest clients bring in most of the income. Serving them still means invoices, contracts, travel bookings and a working laptop. Cut the “trivial” work and the big clients suffer. When you rank tasks by value, trace each important one to the end and count everything it needs.

Grocery sales: heavy buyers matter, and the share varies

In grocery sales, a small group of heavy buyers accounts for most of a typical brand’s revenue, but seldom exactly four fifths of it. The New York University study drew on about 18,000 US households whose shopping A.C. Nielsen tracked with in-home scanners.²

Category averages ran from under two thirds for detergents to close to nine tenths for cigarettes. The share tended to be higher where people bought more often and spent more, and for niche brands; it was lower for larger brands and categories and where more purchases were made on promotion.²

Marketing researchers have long explained this concentration by how often people buy: frequent buyers pile up purchases while occasional buyers add little each.² Think of a café, where the daily regulars are a minority of faces and most of the till.

Work out your own ratio from your own records before you plan around one. A business with a steep curve depends on keeping a few customers; one with a gentle curve depends on reaching many.

Health spending: extreme at the very top, and the top keeps changing

US health spending is extremely concentrated at the very top. In 2022, the 5 percent of people with the highest health-care expenses accounted for about half of all spending, and the bottom half of the population for less than 3 percent, according to the US Agency for Healthcare Research and Quality’s Medical Expenditure Panel Survey.⁷

By the survey’s concentration curve for 2015, the top fifth of spenders accounted for roughly four fifths of spending, close to the classic ratio.⁸ So the same survey can match the classic ratio at one cut and look far more lopsided at another: where you draw the line decides the ratio.

The more useful lesson is about change. Of the people in the top tenth of spenders in 2012, fewer than half were still there in 2013. A small share had died or left the survey population; the rest had moved down the ranking.⁹ A top group measured last year is a snapshot, not a list of fixed members.

The same holds at work. This quarter’s biggest source of rework, or your most demanding client, may not be next quarter’s. Rank again on a schedule instead of acting on an old list.

Does a fifth of your work produce most of your results?

Nobody has measured it. No study we could find tracks what share of one person’s tasks produces what share of their results, so the productivity version of the rule is an extrapolation from sales, defects and income. The publisher’s description of Koch’s book states the claim; it offers no measurement behind it.⁴

Even output compared across people, rather than across tasks, does not follow one shape. A 2017 study by Hyewon Joo, Herman Aguinis and Kyle Bradley compared 229 samples of individual output, from writers and athletes to call-center employees and assemblers, and reports that exponential-tail distributions, rather than pure power laws or the other shapes it tested, were likely the dominant shape in three quarters of them. We read only the abstract.¹⁰

The extrapolation is also harder than it looks. Sales and bug fixes come in one unit that can be counted and sorted; a person’s results mix a signed contract, a colleague helped and a mistake avoided, and some arrive months after the work behind them.

What does transfer are three habits of thought: expect imbalance, measure how much, and check again later. Applying the rule to your hours and your calendar is a separate question from the principle itself, with its own evidence to weigh.

The bottom line

Trust the imbalance, not the ratio. Real data run from under four fifths, as in the average grocery brand, to about half of everything for the top twentieth alone, as in US health spending, and the claim that a fifth of your own effort does most of the work has not been measured. Sort your own numbers, trace what the top few depend on, and sort them again next quarter.