The average isn't the story

The average is one of the most used numbers in business. It's simple, shareable, and familiar. Most people know how to calculate one and most audiences know how to read one. And this makes it feel like a safe choice.

But averages are also one of the most reliably misleading numbers in business. The same average can describe completely different realities, and the version you see depends entirely on how the calculation was set up, and whose perspective it was designed to reflect.

The same number, different stories

An average satisfaction score of 6.5 out of 10 can look encouraging. It can also be the result of half your customers rating you 9 and the other half rating you 4. The average is accurate, but misleading.

An average income trend moving upward can suggest broad prosperity. It can also be driven almost entirely by growth at the top, with little movement for everyone else. Again, the number is correct, but what it implies is not.

Averages flatten differences. They absorb variation and return a single tidy figure. That tidiness is useful when the data is genuinely consistent, but when the underlying distribution is uneven, an average doesn't summarise the data so much as obscure it.

A concrete example: class sizes

Let’s consider a simple example. You want to calculate a university’s average class size. It has 3 classes (one with 20 students, one with 80 students, and one with 140 students).

To calculate the class size average, you likely do the following equation:

(20 + 80 + 140) ÷ 3 = 80 students

Easy right. This is a common way to calculate this metric.

But what if you centred the perspective of each student rather than each class? Each student in the small class experiences a class of 20. Each student in the mid-sized class experiences a class of 80. Each student in the large class experiences a class of 140. Averaged across all 240 students, the equation becomes:

((20×20) + (80×80) + (140×140)) ÷ 240 = 110 students

Both averages are correct; they’re just answering different questions. Averaging across classes describes how resources are distributed. Averaging across students describes what it actually feels like to be at that university. The default choice (averaging across classes rather than people) makes the university appear smaller than most students experience it to be.

When you default to one perspective without realising you’ve made a choice, your analysis answers a question you didn't mean to ask. In this case, "what is the average class size (from the institution’s perspective)?"

The choice you're making without knowing it

This is the deeper issue with averages: they always reflect a perspective, even when that perspective feels invisible. The choice of what to average across (classes, or students) is an analytical decision. So is the choice of what to report.

When that choice goes unexamined, the analysis tends to default to the most institutionally convenient framing. The one that makes things look smoother, more consistent, more manageable than they are. The one that makes variation look like noise.

This matters in organisational data too. An average response time, an average engagement score, an average performance rating—each of these can mask significant variation within the group being measured. But who sits above that average and who sits below it, and why, is often where the real story lives.

What to look at instead

The answer is not to abandon the average, but not to stop there. A few questions worth building into any analysis that uses one:

• How is the data distributed? Are most values clustered near the average, or spread widely? A distribution tells you whether the average is representative or misleading.

• Who does this average reflect? Whose experience is centred in the calculation, and whose is obscured? Changing the unit of analysis (from institution to student, from company to employee, from aggregate to subgroup) can surface a completely different picture.

• Are there meaningful differences within the group? Segmenting by subgroup often reveals patterns the overall average conceals. This is where the insight tends to be.

Why this matters for data storytelling

Every analytical choice is also a communication choice. The average you report, the perspective you centre, the variation you surface or suppress. These shape what your audience understands, and what they decide to do.

When you default to the most convenient framing, you often end up telling the most familiar story rather than the most accurate or neutral one. And familiar stories tend to leave existing patterns undisturbed.

The skill of data storytelling includes knowing which questions to ask before you build anything. This includes whether the average you're about to report is the one that actually answers the question your audience needs answered.

This comes up in the Rogue Penguin Advanced and Applied workshops.

Kat Greenbrook is a data storytelling consultant, author, and workshop facilitator based in Wellington, New Zealand. She is the founder of Rogue Penguin and the author of The Data Storyteller's Handbook.