Does the Pareto Principle Always Hold? Counterexamples and Data Checks
Published:
Last Updated:
Category: Marketing Budget & KPI
Authors: Shusaku Yosa
The Pareto principle can prompt a useful question: are results concentrated in a relatively small number of contributors? It is not a rule that every market or organization must split exactly 80/20. Before applying it, define the item being counted, the outcome, and the period. A weak concentration is a valid finding, not a failed analysis.
Treat 80/20 as a proposition to check
“A few account for much of the outcome” is different from “the top 20% always account for 80%.” Juran’s 1974 account explains the naming history and his attention to the important few. A historical explanation does not demonstrate a fixed distribution in every setting.
Sort customers, products, or other items by a defined outcome and calculate their cumulative share. If the result is dispersed, concentrating exclusively on a few contributors may be the wrong starting point. The purpose is to learn about the data, not to force the data into a familiar slogan.
Keep the two denominators separate
With 10 products, the top two make up 20% of the items. If their revenue is 800 out of total revenue of 1,000, they account for 80% of revenue. The two percentages describe different things and use different denominators. They do not need to add to 100%.
Share | Numerator | Denominator |
|---|---|---|
Share of items | Number of selected products | Number of eligible products |
Share of revenue | Selected products’ revenue | Revenue of eligible products |
Share of gross margin | Selected products’ gross margin | Gross margin of eligible products |
You can also work in the other direction: find how many products are required to reach a selected revenue share. State the threshold as an analytical choice, rather than a universal business boundary.
Compare three distributions with the same total
The following values are hypothetical arithmetic examples, not industry findings. Each set contains 10 items totaling 1,000. The selected two items are always 20% of the items, but their outcome share differs.
Distribution | Values | Top-two outcome share |
|---|---|---|
Equal | Ten items of 100 | 200 / 1,000 = 20% |
Moderate | 300, 200, 100, 100, 100, 50, 50, 40, 30, 30 | 500 / 1,000 = 50% |
Concentrated | 500, 300, and eight items of 25 | 800 / 1,000 = 80% |
Allocating everything to two items in every set would ignore most of the outcome in the equal case. Even the concentrated case does not prove that the next unit of investment belongs with the existing leaders. Additional demand, cost, and capacity still matter.
Check periods, units, and the outcome definition
A large one-time purchase may change a monthly customer ranking. Combining branches into one corporate customer may change it again. Returns, discounts, dormant accounts, missing records, and duplicate identities can all affect the result. Fix these rules before comparing periods.
Revenue rank need not match gross-margin rank. A large account may require substantial support, while a smaller one may be cheaper to serve. If aggregate gross margin is zero or negative, cumulative percentages can behave unexpectedly. Show the amounts and signs rather than interpreting them as an ordinary positive revenue distribution.
A long tail can coexist with concentration
A few popular products may generate a large share of sales while numerous smaller products collectively remain valuable. Those observations are compatible. A long tail is not simply the opposite of concentration. Examine the value and costs of both the head and the tail.
Do not infer an even distribution just because a category sounds commonplace. Purchasing frequency, customer mix, channel, and period can change concentration. Likewise, a long catalog is not proof of profitability: storage, production, and service costs remain relevant. Distribution alone does not select the best operating model.
Do not substitute ant stories for business evidence
Research into ant activity and worker substitution measures something different from customer revenue shares. A Hokkaido University research release from 2016 distinguishes a fatigue-based simulation from observations in actual colonies. It does not establish a fixed human workforce split or a customer-nurturing rule.
- Define the eligible items, period, outcome, and missing-data rules.
- Calculate concentration and compare rankings across suitable periods.
- Check profitability and additional costs before prioritizing action.
- Separate historical concentration from the causal effect of new investment.
See working-ant research and the 2:6:2 story for the evidence boundaries, KPI progress and capacity for recovery decisions, and customer resource priorities for applying concentration without equating revenue with loyalty or incremental value.
