Charts that tell the truth: visualizing retail data

A bar chart lands in the deck with its y-axis starting at 9,500. One bar reaches 9,600 and the other 9,900. On screen, the second bar looks four times taller. The actual difference is 300 units on a base near 10,000. Retail data visualization goes wrong this way constantly, and usually without anyone intending to mislead.

Honest charts aren't about style. They're about making the visual impression match the size of the change.

Why do truncated axes mislead?

Because bar length is how people read bar charts. If the axis starts at zero, a bar twice as long means twice as much. If the axis starts somewhere else, that relationship breaks, and readers still read length.

For bar charts, start the axis at zero. It's a simple rule, and it prevents the most common distortion in business charts.

Line charts are different. A line shows change over time, and starting the axis near the data's range can be appropriate, as long as the axis is clearly labeled. The distortion risk is in the scale. A line that moves from 9,600 to 9,900 on a narrow axis will look like a dramatic climb. Showing the axis values plainly lets readers judge.

Which chart types fit which questions?

Match the chart to the question.

Change over time: a line chart. It shows direction and pace clearly. Use one line per series, and keep the count low so the lines stay readable.

Comparison across categories or stores: a bar chart, sorted by value. Sorting helps readers see the order without hunting.

Parts of a whole: a stacked bar or a simple table often works better than a pie chart, which is hard to read beyond a few slices.

Relationship between two measures: a scatter plot. Each point is a store or an item. It shows whether the measures move together, and how tightly.

The data visualization field has deep literature on these choices. For most retail work, these four cover nearly every need.

What other traps should you avoid?

Dual axes. A chart with two y-axes, one for sales and one for price, lets the designer make the lines cross wherever they like by adjusting each scale. Readers see a relationship that the scales created. Two separate charts, stacked, are usually clearer.

Cumulative totals. A running total always goes up. It hides slowdowns, since the line keeps rising even when each period adds less. Show period values if the question is about pace.

Unequal time intervals. If the x-axis mixes weeks and months, or skips periods, the line's slope stops meaning anything. Keep intervals consistent, and mark gaps.

And too much at once. A chart with twelve lines and a legend nobody can match to colors doesn't inform. Split it, or highlight the one or two lines that matter and gray out the rest.

How should charts be labeled?

Completely. Title the chart with what it shows and, ideally, the takeaway. Label both axes with units. State the period, the store basis, such as same-store or all stores, and the data source.

Note anything unusual. If a period includes a holiday shift, a store count change or a data gap, a short note on the chart prevents misreading.

And keep the underlying numbers available. A table beside the chart, or in an appendix, lets a skeptical reader check. The monthly reports from NRS Insights present category trends over time, and past releases are on the news page.