Five friends get weekly allowances of $5, $6, $7, $8 and $9. The mean and the median are both $7. Now the friend who gets $9 is given $50 instead. The median is still $7, the third value in order. The mean becomes (5+6+7+8+50)÷5=76÷5=15.2, so $15.20, which is more than four of the five friends receive.
The reason is in how each center is built. The mean uses the size of every value: the extra $41 is shared among all five friends, adding 41÷5=8.2 dollars to the mean. The median uses only the order: as long as the top value stays on top, its size makes no difference. So the median resists outliers and the mean does not, and the bigger the extreme value, the further apart the two centers move.
A gap between mean and median does not always come from one outlier. A whole data set can lean to one side with a long tail, like house prices, where many homes are moderately priced and a few are very expensive: the mean then sits above the median (skewed right). A long tail toward low values, such as scores on an easy test, puts the mean below the median. When the data are roughly symmetric, the two are close, and either one describes the center.
Think of it like this. The mean behaves like the balance point of a seesaw loaded with children: slide one child far out along the board and the pivot must move toward them to keep it level. The median behaves like the middle child in a queue: move the child at the end anywhere you like and the middle child stays the middle child.
Core idea. The mean uses every value's size, so an extreme value drags it; the median uses only position, so it stays put. A large gap between them signals skewed data and tells you to report the median.