Average, typical, and the odd one out.

The mean adds everything up, so one big value drags it along. The median is the middle value once they’re sorted, so it only cares about order. Report the median for typical order values, salaries or delivery times, where a few big ones are normal; use the mean when every rupee counts, as in total revenue ÷ orders.

When the mean sits well above the median, a long right tail is pulling it, and that tail is often the interesting part. The monthly charge, by contrast, is balanced: mean ₹501, median ₹499. Mostly-zero numbers are often best summed up as a share: “65% made no complaint”.

Percentiles describe spread in the data’s own units, which a manager understands at once: “most customers pay between ₹309 and ₹695”. The standard deviation (₹150 here) sums up spread in one number; for bell-shaped data, about 95% of values sit within two of them of the mean.

Before any summary, check the range and the blanks: the minimum, the maximum and the count of missing values (245 charges are missing here). An outlier can be a rare real case worth studying or an error worth fixing, and you decide which by looking, not by deleting it automatically. A common rule flags values more than 1.5 interquartile ranges beyond the quartiles.

Big groups produce big counts; rates make groups comparable. Divide by each group’s size before comparing cities, months or campaigns, and say how big each group is, because a rate from 31 customers wobbles far more than one from 3,249.