A chart is not a decoration for data. It is an argument about what the data means. Draw the same dataset as a bar chart, a line chart, and a pie chart, and you get three different arguments — one of which answers the question you actually asked, and two of which answer questions you did not ask, or worse, suggest an answer that is not there. Choosing a chart type is therefore not a styling decision. It is the decision that determines whether the reader understands the data or is misled by it.
Chart Builder builds bar, line, pie, scatter, and histogram charts from CSV data in the browser, with aggregation, sorting, and accessible colour palettes, exporting as PNG or SVG. This post takes one fictional dataset and draws it three ways, showing which chart answers which question — and where each one misleads.
The dataset: a fictional shop's monthly sales
A fictional shop, Example Home Goods, records monthly sales across four product categories for one year. For this post, we focus on a single summary question first: how did total sales change month by month? The monthly totals, invented for illustration, are: January £12,000, February £11,000, March £14,000, April £13,500, May £16,000, June £15,000, July £18,000, August £17,500, September £15,500, October £19,000, November £21,000, December £24,000. The story in the numbers is a rising trend with small ups and downs, ending in a strong December.
That story has a natural chart — but the point of this post is to see what happens when the natural choice is not the one used.
The line chart: the trend, made visible
A line chart connects the monthly totals in time order, and the eye reads the slope. The rises and dips are visible, the overall upward direction is unmistakable, and December's peak stands out as the end of a climb rather than an isolated spike. For a question about change over time, the line chart is the right tool, because time is continuous and the line's shape is the trend itself.
Its weakness is precision: a reader can see that July beat June, but cannot easily read exact values off the line. That is acceptable when the question is about direction and shape. If the question were "which month sold exactly how much," a table or a labelled bar chart would serve better. Every chart trades something away; the line chart trades exactness for trend, and for a time question that is the right trade.
The bar chart: the comparison, made exact
A bar chart of the same twelve months shows each month as a separate bar, and the eye compares heights. December towers over February; May beats April; the autumn climb is a staircase. For a question like "which months were strongest, and by how much," bars are excellent, because each value has its own visual weight and the comparison is direct.
Bars also tempt a specific dishonesty: truncating the vertical axis. If the axis starts at £10,000 instead of zero, February's £11,000 bar looks tiny next to December's £24,000, exaggerating the difference several-fold. A bar chart's honesty depends on its axis starting at zero, because bar length is the value. A line chart can use a truncated axis more defensibly, since it shows shape rather than length — but for bars, a non-zero baseline is a distortion, not a design choice. Always check the axis before trusting a bar chart, including your own.
The pie chart: the wrong question, answered confidently
Now draw the same twelve months as a pie chart: twelve slices, each month's share of the year's total. The chart is technically correct — the slices do sum to the year — but it answers a question nobody asked. Nobody needs to know that December was 11% of the year's sales and February 5%; the interesting fact is the trend, which a pie destroys by removing time order and turning a sequence into a set of shares. Twelve thin slices are also nearly unreadable: adjacent months blur together, labels crowd, and the eye cannot compare slices that are not next to each other.
Pie charts have a legitimate use, narrowly defined: a small number of categories (two to five) where the question is genuinely about shares of a whole — for example, four product categories' shares of annual revenue. Even there, a bar chart of the same shares is usually easier to read. The rule is simple: if the question is "how has this changed over time," do not use a pie. If the question is "what share is each part of the whole," a pie is acceptable for few categories — and a bar chart remains a safe alternative.
Worked example: the category question, answered twice
Return to Example Home Goods with a second question: what share of annual revenue came from each of four categories? The fictional annual totals are: Kitchen £96,000, Bedroom £72,000, Garden £54,000, Bathroom £42,000 — a total of £264,000, or shares of roughly 36%, 27%, 20%, and 16%.
As a pie chart, this works: four slices, clearly different sizes, with labels, answering "what share" at a glance. As a bar chart, it also works — and arguably better, because the bars can be sorted largest to smallest and labelled with exact values, making the Kitchen-versus-Bedroom gap precise rather than approximate. The lesson is not that pies are forbidden, but that they are rarely the best answer even to their own question. When in doubt, the bar chart is the safer default: it handles comparisons, shares (with labelled values), and rankings, and it fails gracefully where a pie fails confusingly.
Honest limitations
Chart Builder draws the chart you ask for; it does not judge whether that chart type fits your question. A misleading chart made with correct data is still misleading, and aggregation choices — what is grouped, what is averaged, what is left out — shape the result as much as the chart type does. Always state the question before choosing the chart, check axes (especially bar baselines), label values where precision matters, and remember that a chart built from uncleaned data inherits every problem in that data — which is why the workflow runs through CSV Cleaning Studio first. For preparing surrounding images, Image Optimizer can resize and compress exports for the web. The background guide is how to choose the right chart.
FAQs
Should I use a bar chart, line chart, or pie chart?
It depends on the question. Use a line chart for change over time, a bar chart for comparing values or ranking categories, and a pie chart only for shares of a whole across a small number of categories — and even then, a bar chart is usually clearer.
Why is a pie chart wrong for monthly sales over a year?
Because it removes time order and turns a trend into shares. Twelve thin slices are hard to read and harder to compare, and the question "how did sales change" is lost. A line chart shows the trend; a bar chart shows the monthly comparisons.
Must a bar chart's axis start at zero?
Yes, for honest bars. Bar length represents value, so a truncated axis exaggerates differences. A line chart, which shows shape rather than length, can use a different baseline more defensibly — but bars should start at zero.
Can a chart mislead even with correct data?
Yes. The wrong chart type, a truncated axis, an unlabelled aggregation, or a missing category can all mislead while every plotted number is technically correct. Check the chart type, the axis, and the grouping before trusting the picture.
human review pending client designation