General

A practical way to report business metrics when your data is incomplete, delayed, or built on assumptions, without pretending you have more certainty than you do.

Skewness is why your average order looks bigger than any real order. Here is what it means, how to spot it with mean versus median, and what to report instead.

The normal distribution is the bell curve your stats shortcuts secretly assume. Here is what it is, why business data rarely fits, and how to check.

Why the customers who talk to you most and buy the most can quietly bias your product and pricing decisions, and how to catch it.

Percentiles read the tail your average hides, with no bell shape assumed. Here is what p50, p90, and p95 mean, when to trust them, and the traps.

Standard deviation is the single number for how far your values sit from their average. Here is what it measures, the rough rule you can lean on, and the traps.

A plain-English guide to picking an analytics baseline that holds up under scrutiny, with the traps solo operators fall into and how to avoid them.

A confidence interval is the honest way to show a number as a range instead of a single false point. Here is what the range and its confidence level mean, and how to read yours.

Sample size is the count of events behind a number. Here is why a small one is mostly luck, what really drives how much you need, and the honest answer for a business of one.

My active customer count jumped 19 percent overnight because I had edited one line of a query eleven days earlier and forgotten. A metric definition with no effective date turns your own history into a rumour.