What it means
A symmetrical distribution is balanced around its centre, so the mean, median and mode all sit at roughly the same point. An asymmetrical, or skewed, distribution has a longer tail on one side, and that tail drags the mean towards it.
Right skew, meaning a long tail of unusually high values, is far more common in business than left skew. Averages are the default summary in almost every management report, and on skewed data they mislead.
A handful of enormous contracts can lift average deal size well above what a typical customer actually spends, which produces sales targets and cash forecasts built around a number nobody hits. The quickest diagnostic is to compare the mean with the median: if the mean is materially higher, the data is right skewed.
A histogram or a simple quartile table confirms the shape, and a skewness coefficient puts a single number on how severe the imbalance is. Once you know a series is skewed, report the median alongside the mean and show a range rather than one figure.
Percentiles are frequently more useful than averages, because the 90th percentile of order value tells an operations team what capacity to plan for in a way the mean never will. Skew is not an error waiting to be corrected.
It usually reflects something real about the business, such as a few very large accounts or occasional very long payment delays, and modelling it honestly beats pretending the data is neatly bell shaped.
In practice
Real-world examples.
Example
A subscription business reports average customer lifetime value of $4,100, but the median is $1,300 because a small enterprise segment dominates the total. Marketing had been bidding for new customers at a cost based on the mean and quietly losing money on the bulk of the base.
Example
A logistics operator finds that average delivery delay is 40 minutes while the median is 9 minutes, since a handful of failed routes run into many hours. The team stops chasing the average and starts targeting the 95th percentile, which is what customers actually complain about.
Example
A payroll review shows a mean salary comfortably above the median because two senior hires sit far above everyone else. The finance director presents quartiles instead of a single average, and the pay review conversation becomes far more grounded.
Formula
Calculation
Pearson's Second Coefficient of Skewness = 3 x (Mean - Median) / Standard Deviation
A consultancy completed 200 projects last year. The mean project fee was $92,000, the median was $68,000 and the standard deviation was $60,000.
Skewness = 3 x ($92,000 - $68,000) / $60,000 = 3 x $24,000 / $60,000 = $72,000 / $60,000 = 1.2
A coefficient of 1.2 is strongly positive, confirming heavy right skew: a small number of very large engagements is pulling the mean upwards. Total revenue is still mean x volume, so $92,000 x 200 = $18,400,000 is the correct total. What is wrong is describing $92,000 as a typical project, because at the median the equivalent figure would be $68,000 x 200 = $13,600,000, a gap of $4,800,000. Sales leaders who set individual quotas from the mean are asking their team to win projects that occur only in the top part of the range.Case study
Seen in the real world.
Thornbury Analytics is a fictional market research firm invented to illustrate the danger of averages. Its board approved a growth plan built on the statement that the average client contract was worth $92,000 a year and that adding fifteen clients would therefore add roughly $1.4 million of revenue.
Nine months later the firm had won seventeen new clients and added only about $800,000. A junior analyst plotted the contract values and found the distribution was heavily right skewed, with a median of $68,000 and three legacy accounts above $400,000 propping up the mean. New business was landing near the median, exactly as it always had.
In this illustrative example the fix was presentational as much as statistical. Thornbury rebuilt its pipeline model around median contract value, added an explicit line for large deals treated as separate opportunities, and reported both mean and median on every board pack. Forecast accuracy improved not because the maths got cleverer, but because the shape of the data was finally visible.
Watch out
Common mistakes.
- Quoting the mean as the typical value without checking the median. On skewed data the mean can sit well above the experience of most customers, employees or transactions.
- Deleting high values as outliers to make the data look tidy. Those large accounts are often the most profitable part of the business and deserve separate analysis rather than removal.
- Applying rules of thumb built for a normal distribution, such as expecting about two thirds of values within one standard deviation. Those rules break down badly once a series is strongly skewed.
Questions
People also ask.
How do I tell which way data is skewed?
Compare the mean and median: a mean above the median points to right skew, and a mean below it points to left skew.
Is a skewness coefficient of 1.2 large?
Yes, values beyond about 1 in absolute terms are generally treated as strong skew, while values between roughly -0.5 and 0.5 suggest something close to symmetry.
Should skewed data always be transformed?
No, transformation helps some statistical techniques but makes results harder to explain, so for ordinary business reporting it is usually better to show medians, ranges and percentiles.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%Related