When Numbers Mislead

A company announced that its average delivery time had fallen from five days to three. The headline sounded impressive. However, an average alone cannot reveal whether the improvement was shared by all customers.

Suppose most orders arrive the next day, while a small group takes much longer. The average may look acceptable even though those customers experience serious delays. A different measure, such as the proportion of orders delivered within three days, can answer a different question. Neither measure is automatically the right one for every situation.

Numbers become useful when we connect them to the decision we need to make. A customer ordering a birthday gift may care more about reliability than average speed. Before celebrating a statistic, we should ask what it includes, what it hides, and whose experience it represents.