Standard Deviation

Analytics

Also: SD · Std Dev

Standard deviation = how far, on average, each result sits from the mean
What it measuresSpread of results around the average
Low value meansResults cluster close together
High value meansResults are scattered widely
Watch forSmall samples inflate it

Quick definition

Standard deviation is a measure of how spread out a set of numbers is around its average. A low standard deviation means most results sit close to the mean. A high one means results are scattered widely, which affects how much you should trust any single average.

What it actually means

Picture two delivery drivers with the same average delivery time of thirty minutes. One arrives between twenty-eight and thirty-two minutes every time. The other swings anywhere from ten minutes to an hour. Same average, wildly different reliability. Standard deviation is the number that tells you which driver you're dealing with.

In marketing, standard deviation shows up wherever you're tempted to trust a single average. Conversion rate across landing pages. Order value across customer segments. Response time across A/B testing variants. If the spread is tight, the average is a fair summary of reality. If the spread is wide, the average is hiding a lot of noise.

This matters most in experimentation. A/B testing results with a small sample size and high standard deviation can show a big lift that's actually just noise dressed up as a pattern. Statistical significance calculations use standard deviation directly to work out whether a difference between two groups is real or coincidental. Without it, you're reading tea leaves and calling it data.

An average without a standard deviation is a story with no sense of how reliable the narrator is.

How to calculate it

Standard deviation = square root of the average squared distance of each value from the mean

Worked example. Five landing pages convert at 2%, 3%, 3%, 4% and 8%. The mean is 4%. Most pages sit within one point of the mean, but the 8% outlier pulls the spread wide. The standard deviation captures that the average of 4% doesn't represent most of your pages well.

The Australian context

Australian businesses often run experiments on smaller traffic volumes than US or UK equivalents because the addressable market is smaller. Smaller samples naturally produce higher standard deviation and noisier results. That means Australian marketers need to run A/B testing for longer, or accept wider margins of error, before calling a winner.

Where people get this wrong

Reporting an average without mentioning the spread.Two data sets can share an identical average and tell completely different stories. Without standard deviation, the average alone hides how much you can trust it.
Calling an A/B testing result significant with a tiny sample.Small samples produce inflated standard deviation, which makes random noise look like a real lift. Statistical significance needs both sample size and spread accounted for.
Assuming a high standard deviation means bad data.A wide spread often reflects real variation in customer behaviour, not a measurement error. The fix is usually segmentation, not distrust of the numbers.

Related terms

Common questions

What does a high standard deviation mean in marketing data?

It means your results are widely scattered rather than clustered near the average. A high standard deviation on conversion rate, for example, suggests some pages or segments perform very differently from others, and the average alone won't tell you which.

Why does standard deviation matter for A/B testing?

Statistical significance calculations use standard deviation to judge whether a difference between two groups is a real effect or random noise. High spread combined with a small sample makes it easy to mistake a lucky result for a genuine winner.

How is standard deviation different from average?

The average tells you the central value of a data set. Standard deviation tells you how far individual results typically sit from that central value. You need both to understand a data set. The average alone can mislead.

What's a good standard deviation for marketing metrics?

There's no universal good number. It depends entirely on the metric and the scale it's measured on. What matters is comparing the standard deviation to the mean itself. A spread that's small relative to the average signals a reliable, consistent metric.

Debrief

Get the next one

No spam. No fluff. Just the next article, straight to your inbox.

Keep exploring

About New Rebellion

New Rebellion is a marketing intelligence consultancy. We build tools, score Australian businesses on how their marketing actually performs, and publish Debrief every day. This dictionary is part of how we work in the open.

How we think →