Skewness measures how asymmetrical a distribution is.
Distributions with skewness close to zero are close to symmetrical. Distributions with skewness further from zero are more asymmetrical.
Distributions with a negative skewness are "left skewed". This means they have a longer tail on the left hand side. Distributions with a positive skewness are "right skewed" and have a longer tail on the right-hand side.
When To Use Skewness
Skewness is useful when you want to measure how symmetric a distribution is. A highly skewed distribution will be more influenced by values on one of the tails than the other.
Calculating skewness and kurtosis will tell you about how symmetrical a distribution is and how much of it is in the tails.
How To Calculate Skewness
Pearson's Median Skewness can be calculated as 3 * (mean - median) / std. Where std is the distribution's standard deviation.
The numerator (mean - median) means that distributions with a mean lower than the median will have negative skewness. And distributions with a mean higher than the median will have positive skewness.
Dividing by the standard deviation normalizes the skewness measure to the overall variation in the data. A distribution with a (mean - median) of 5 and a standard deviation of 100 probably looks pretty symmetrical. While another distribution with a (mean - median) of 5 and a standard deviation of 3 probably looks right skew (a longer tail of numbers on the right-hand side).