Pattern/Mental Model/No. 0721
Pareto Principle
The Pareto principle, or 80/20 rule, is a heuristic stating that a small share of causes often produces a large share of results. Named by Joseph Juran after Vilfredo Pareto’s work on income distribution, it is used in quality management; the exact proportions vary.
Also called Pareto Principle (80/20 Rule) · 80/20 Rule · Law of the Vital Few
- Evidence
- Useful, modest evidence
- Read
- 6 min
- Links
- 8 connections
01You've seen this when…
- in life
Your recipe folder holds 90 dishes. Most dinners come from the same six.
- at work
The team has 40 customers, but five account for most support hours. Everyone still gets the same monthly check-in.
- out in the world
A city maps a year’s serious road injuries. A handful of junctions keeps appearing, while most junctions have none.
02The idea
A long list can make every item look equally important. Sort the items by what they contribute, and a different picture often appears: a few account for much of the total, followed by many that each contribute little.
The Pareto principle is the rule of thumb that a minority of inputs often accounts for a large share of outputs. A few customers generate much of the revenue. A few recurring faults generate much of the repair work. A few expenses absorb much of the budget.
The familiar 80/20 split illustrates a pattern whose proportions vary. Your data might show 10% of causes producing 60% of the result, or 30% producing 90%. The two percentages describe different totals, so their sum can differ from 100.
Its practical message is to look for concentration before spreading attention evenly. Whether doing 20% of the work will deliver 80% of the result depends on the case.
It is also separate from Pareto efficiency, which concerns whether someone can be made better off without making someone else worse off.
03Why it happens
Several mechanisms can produce uneven contributions. None guarantees an 80/20 split.
- The contributors differ in size. A customer with 2,000 employees can create far more revenue or support work than one with ten. Counting each as one customer hides that difference.
- One recurring cause can create many incidents. A faulty component used across several products can generate hundreds of returns. Ten unrelated faults might produce only a handful each.
- Success can feed further success. A popular product gets more visibility, which brings more buyers and more visibility. This kind of reinforcing feedback can concentrate results.
- Some outcomes vary enormously. Book sales, business sizes and personal incomes can span wide ranges. A few very large values can dominate the sum.
Some concentrated datasets follow a power-law distribution. Establishing that particular mathematical shape takes evidence beyond concentration alone. You can use the principle as a rule of thumb while leaving claims about a universal law open.
04A worked example
Imagine a pump manufacturer reviewing 1,000 returned units. Inspectors assign each unit one primary fault. With 480 returns attributed to loose electrical connectors and 300 to leaking seals, eight other categories account for the remaining 220. This is an illustrative, hypothetical factory case.
What it looks like Ten fault categories competing for engineering time. Giving each category a similar budget seems balanced.
What’s actually going on Two of the ten categories account for 78% of returns. Correcting those recurring faults could reduce far more repair work than dividing effort evenly. The best investment remains uncertain from the table alone. Connector failures might be expensive to prevent, and a rare fault among the other categories might pose a serious safety risk.
What would have helped After ranking faults by frequency, the team could assess severity and compare repair costs with the cost of prevention. The team could test a connector change on a limited production run and compare subsequent failure rates. If returns fall, it has evidence for expanding the fix that goes beyond a chart resembling 80/20.
05How to spot it
These signs call for investigation before drawing conclusions. Memory favors vivid cases, and grouping choices can make concentration look stronger or weaker.
06What to do about it
- Define the result you care about. Revenue, profit, complaints and customer harm are different outcomes. A customer who dominates sales may contribute little profit. Choose the measure before ranking contributors.
- Build a ranked list from actual data. Choose a meaningful period. Total each contributor’s share and sort from largest to smallest. Calculate the cumulative share as you move down the list. Choose the cutoff from the data, with 20% serving only as a reference.
- Check what the counts leave out. A busy junction may have many crashes partly because many vehicles pass through it. Total harm and harm per journey answer different questions. Missing records and measurement error can also distort the ranking.
- Rank possible changes by their expected payoff. The largest contributor may be harder to improve than smaller ones. Use marginal analysis: compare the expected benefit of the next action with its cost and the opportunity cost of alternatives.
- Test the leading opportunity. Make one bounded change and measure the result. Use what you learn to update the ranking. A large category points to where to investigate; the proposed fix still needs evidence of effectiveness.
- Protect what the ranking cannot value. Meet legal obligations and minimum service standards while keeping safety controls in place. Turning one concentrated metric into the whole goal invites Goodhart’s law.
07Where it doesn’t tell you what to cut
A low-frequency event can still be important. An emergency shutoff may never get used and still be essential. A small customer may deserve support under a contract. A rarely performed maintenance task may prevent an expensive failure.
Finding a bottleneck requires examining what constrains the system, beyond measuring concentration. The activity producing most visible output may depend on a quieter activity that limits the whole system. Removing the quiet one can damage everything else.
The pattern can also shift after you act. Once the largest defect is fixed, another becomes the leading source of trouble. Repeating the same intervention may yield less.
Finally, an 80/20 pattern across the whole list does not imply another 80/20 pattern within its top fifth. The shortcut that 4% of inputs must produce 64% of outputs requires an additional assumption, not simple arithmetic.
08Roots
In the 1890s, Vilfredo Pareto was teaching economics in Lausanne and comparing income data drawn from tax records. The striking feature was the upper end: a relatively small number of people had very large incomes. He explored regularities in those distributions in his Cours d’économie politique. His work concerned income distributions. A universal instruction to discard four-fifths of a to-do list would go beyond those findings.
Decades later, quality-management engineer Joseph Juran needed a way to direct managers’ attention. A factory could record many kinds of defects, yet a few sources of trouble could account for much of the loss. Treating every category as equally urgent wasted effort.
Juran attached Pareto’s name to this broader pattern and brought it into quality management, including his 1951 Quality-Control Handbook. Sorting problems by their contribution made the distinction between the vital few and the rest visible and actionable.
In a later article, Juran acknowledged that extending Pareto’s work into this general management principle had been his own contribution. The memorable 80/20 shorthand traveled well beyond factories. It became advice about sales, spending and productivity—useful when it prompts measurement, misleading when the memorable numbers replace it.
09How solid is this?
Concentrated contributions are well documented in income distributions and quality-management applications. The exact proportions vary, with 80/20 serving as a shorthand. Finding the intervention with the best return requires evidence beyond concentration alone.
10Connections
- Often confused withPareto Efficiency, Power-Law Distribution
- See also Marginal Analysis, Opportunity Cost, Bottleneck, Reinforcing Feedback, Measurement Error, Goodhart’s Law
11Origin and sources
Vilfredo Pareto studied unequal income distributions in the 1890s. Joseph M. Juran named and extended the Pareto principle as a quality-management heuristic in the mid-20th century.
- [1]Pareto, V. (1896–1897). Cours d'économie politique. F. Rouge.
- [2]Juran, J. M. (Ed.). (1951). Quality-Control Handbook. McGraw-Hill.
- [3]Juran, J. M. (1975). The Non-Pareto Principle; Mea Culpa. Quality Progress, 8(5), 8–9.
- [4]Newman, M. E. J. (2005). Power laws, Pareto distributions and Zipf's law. Contemporary Physics, 46(5), 323–351.
Suggest an edit· Updated 2026-10-02