Why can't the average be used to calculate discounts in skewed LTV distribution?

The mean is sensitive to outliers and skewed distributions. In the case of LTV (Lifetime Value) with a skewed distribution, where there are a few very large values, the mean will be inflated and not reflect the typical user behavior.

Therefore, it is better to use the median or other outlier-resistant metrics for calculating discounts, so that the discount is fair and reflects the actual value of most customers.