What it means
Finding the median is mechanical. Sort the values, and if there is an odd number take the middle one, while with an even number take the average of the two in the middle.
The median matters because so much business data is skewed rather than evenly distributed. Customer spend, invoice size, time to resolve a support ticket and employee pay all tend to have a long tail of large values that pulls the mean upward and away from the typical case.
This is why the median is the standard measure in pay reporting, property statistics and household income figures. Quoting a mean salary in a company with two very highly paid founders describes almost nobody, while the median describes the person in the middle of the payroll.
The trade off is that the median discards information. It ignores the size of the values on either side, so two very different data sets can share a median, and it cannot be combined arithmetically the way means can be added and weighted.
The practical approach is to use both. Reporting the median alongside the mean immediately shows whether the data is skewed, and adding quartiles or a range tells the reader how widely the values are spread around that middle point.
In practice
Real-world examples
Example 1
A software company reports a median support resolution time of 4 hours against a mean of 11 hours. The gap is caused by a small number of complex cases running for days, so the service level agreement is written around the median with a separate escalation target.
Example 2
An estate agency quotes median sold prices by postcode rather than mean prices. One $4,000,000 sale in a street of $400,000 houses would distort the mean badly, while the median stays representative of what buyers actually pay.
Example 3
A payroll team calculates the median salary as part of annual pay gap reporting. The figure of $46,000 sits well below the mean of $61,000, and the difference itself becomes a talking point about the shape of the pay structure.
Think of it
Imagine five people in a room where four earn 30,000 pounds and one is a billionaire. The average income is millions, but the median income is 30,000 pounds, which actually describes the people in the room.
Formula
Calculation
Median = middle value of the sorted list (odd count), or the average of the two middle values (even count)
A retail group reviews monthly sales at its seven stores: $24,000, $94,000, $18,000, $31,000, $27,000, $44,000 and $21,000.
Sorted: $18,000, $21,000, $24,000, $27,000, $31,000, $44,000, $94,000
With seven values the middle one is the fourth, so the median is $27,000.
The mean tells a different story. The total is $18,000 + $21,000 + $24,000 + $27,000 + $31,000 + $44,000 + $94,000 = $259,000, and $259,000 / 7 = $37,000, which is higher than five of the seven stores actually achieved because the flagship at $94,000 pulls it up. If an eighth store opens and sells $34,000, the sorted list becomes $18,000, $21,000, $24,000, $27,000, $31,000, $34,000, $44,000, $94,000, and with an even count the median is the average of the fourth and fifth values: ($27,000 + $31,000) / 2 = $29,000.Case study
Seen in the real world
The following is an illustrative and entirely fictional example. Halstow Foods, an invented wholesale distributor, set its sales team a target based on mean order value of $2,400 and could not work out why nobody was hitting it.
The analytics lead pulled a month of orders. Of 1,000 orders, 940 fell between $300 and $1,500, while 60 orders to three national accounts ranged from $18,000 to $45,000. Those 60 orders alone accounted for a large share of the total, and the median order value was $780, less than a third of the mean the targets were built on.
Halstow rebuilt its reporting in this illustrative account. Field sales targets moved to median order value with a separate national accounts track, the commission scheme was split accordingly, and the fictional board began receiving both figures on the monthly pack. Median order value rose from $780 to $910 over two quarters as reps focused on realistic upselling, and the sales director noted that the previous target had been demoralising precisely because it described a customer almost none of the team ever dealt with.
Watch out
Common mistakes
- Using the mean by default for skewed data such as salaries, order values or resolution times, where the median describes the typical case far better.
- Trying to combine medians from separate groups by averaging them, which does not produce the median of the combined set.
- Reporting a median from a very small sample and treating it as stable, when a handful of values can move it sharply.
Questions
People also ask
When is the mean better than the median?
When the data is roughly symmetrical, or when you need a figure that multiplies back into a total, such as average cost per unit across a whole production run.
What does it mean if the mean is much higher than the median?
The data is skewed to the right, meaning a minority of large values are pulling the average up, which is normal for income, spend and duration data.
Is the median the same as the 50th percentile?
Yes, they are two names for the same point, the value below which half the observations fall.