Two depots, both averaging 42 orders a day. Same number, same KPI tile — and completely different businesses: one does 40-to-45 every day like a metronome; the other swings between 5 and 90 and merely averages out at 42. Staffing, stock, stress — everything about running them differs, and the average is constitutionally incapable of saying so.

What says so is the distribution — how often each range of values occurs — and its picture, the histogram: values banded into bins along the bottom, frequency as bars. It’s the chart people skip because it doesn’t answer “how much?”; it answers the better question underneath — “what’s typical, how spread out, and what’s lurking in the tails?”

Three ways to draw one

Fastest: select the column, Insert → Statistic Chart → Histogram. Excel bins automatically; right-click the axis to set bin width to something a human thinks in (£10, 5 orders, one week). Done in a minute, fine for a look.

Most control: count the bins yourself with the banding machinery you already own — a small bin Table (From, Label) and one COUNTIFS per row (>= this bin, < the next), charted as an honest bar chart. More work, but the bins are data, the counts feed checks (bin counts must sum to COUNT(all)), and the whole thing refreshes with the workbook. (The modern shortcut: =FREQUENCY(values, bin_tops) spills every count at once.)

Most flexible: a pivot with the value field grouped — the number-grouping trick — which adds slicers for free: the distribution by region, one click each.

the metronome — mean 42 the rollercoaster — mean 42 same red mean line, opposite businesses — the shape was the story all along
Two columns, one average, no resemblance. Every decision that depends on "typical" or "worst case" lives in the shape, not the mean.

Reading the shape

Four things to look for, each with a decision attached:

  • Where’s the bulk? The typical case — often not the mean. Skewed data (money, almost always) drags the mean toward the tail; the median tells the truer centre, and the histogram shows you why they disagree.
  • How wide? Spread is risk. The metronome depot staffs to 45; the rollercoaster needs a buffer policy — same instinct, business edition.
  • One hump or two? Two humps means two populations wearing one column — weekday and weekend orders, two customer types. Split them (a slicer or a FILTER) before averaging anything, or every summary statistic blends two truths into one fiction.
  • What’s in the tails? Outliers — sometimes data errors the checks row should catch (a £84,000 coffee), sometimes the most important rows in the file (the whale customer). The histogram won’t say which; it says look here.

The craft note: bin width is an editorial choice. Too wide flattens the story; too narrow shatters it into noise. Try two or three widths — it’s one setting — and, as with every chart, pick the one that shows the shape honestly rather than dramatically.

Averages are the summary you quote. Shapes are the thing you should look at before quoting it — and now the look costs one minute and one chart.