Your data is skewed (e.g., revenue, deal sizes, salaries, customer LTV) - You want to represent the "typical" case, not the aggregate - Outliers exist and you do not want them to distort the picture