Chart Maker

You edit the data, not a form

The difference from most free chart makers, and it changes how the thing feels to use.

The usual pattern online is a form: type your labels into one box, your values into another, choose a type, press Generate, look at the result, go back and change something, press Generate again. It is slow enough that people stop experimenting.

Here there is a data grid you edit directly and the chart re-renders on every change. Fix a number and the bar moves. Rename a label and the axis updates. Switch from bar to line and the same data redraws instantly.

That makes two useful things easy. Trying chart types costs nothing, so you can see your data as a bar chart and a line chart and decide rather than guessing. And spotting a typo is immediate — a mistyped value shows up as a bar going through the roof, which is far more obvious than a wrong number in a form.

Which chart type to use

The choice that determines whether your chart communicates anything, and the commonest place people go wrong.

Bar chart — comparing amounts across categories. Sales by region, votes by candidate, responses by option. This is the right answer far more often than people think, and it is almost never the wrong one. If you are unsure, use a bar chart.

Line chart — a value changing over time. Months, years, days, readings. The key rule: only use a line when the horizontal axis is genuinely continuous. A line connecting sales in four unrelated cities implies a progression that does not exist.

Pie chart — parts of a single whole, and only then. Two hard conditions: the slices must add up to 100% of something real, and there should be no more than about five. Beyond that nobody can compare them, because people judge angles badly — which is why a bar chart beats a pie chart for almost every comparison. If your slices do not add up to a whole, a pie chart is simply wrong.

Doughnut chart — the same as a pie with a hole, so the same rules apply. It reads slightly better because the eye compares arc lengths rather than angles, and the middle is somewhere to put a total.

Radar chart — comparing a few things across several shared dimensions. Skill assessments, product feature comparisons, survey profiles. Effective with two or three series and unreadable with six.

Polar area chart — cyclical data where the categories wrap round, such as months of a year or directions of a compass. Unusual, and genuinely good for the narrow case it fits.

Making a chart that is actually readable

A handful of things that separate a chart people understand from one they squint at.

Start bar charts at zero. Not a stylistic preference — a truncated axis exaggerates differences and is the single most common way charts mislead, usually accidentally. A bar chart's whole meaning is that bar length is proportional to value, and that breaks the moment the axis does not start at zero. Line charts are allowed more latitude, because they show change rather than magnitude.

Sort your bars. Largest to smallest is almost always more useful than alphabetical, because the order itself tells the reader something. Leave the order alone only when it means something already, such as months or sizes.

Label the axis with its units. "Revenue" tells the reader nothing; "Revenue (thousands, USD)" tells them everything. This is the most frequently omitted thing on a chart and the most frequently asked about.

Use colour to mean something, or not at all. A different colour per bar in a single-series chart is decoration, and it implies the categories differ in some way they do not. One colour for all the bars, with a second colour to highlight the one you are making a point about, is stronger and looks more deliberate. Color Palette Generator will generate a coherent set if you need several.

Drop the legend when you have one series. If the axis labels already say what each bar is, a legend is another thing to read for no information.

Write a title that states the finding, not the subject. "Sales by Quarter" describes the chart; "Q4 sales doubled after the price change" tells the reader what to notice. In a report the second is almost always more useful.

Practical notes

The chart downloads as a PNG image, which drops into a document, a slide, an email or a post directly. There is no vector or SVG export, so if you need a chart that scales for large-format print, you will want a spreadsheet or a charting application instead.

Two things follow from PNG output. Build the chart at the size you need rather than enlarging it afterwards, because enlarging a PNG chart softens the text and the lines. And check the text is readable at final size — a chart that is legible on your screen can be unreadable projected or printed small.

The chart is not saved anywhere, so it exists only while the tab is open. Download it before you navigate away, and keep your numbers somewhere of your own if you might need to rebuild it — there is nothing to come back to.

Everything runs in your browser and nothing is transmitted, which is worth having: the numbers people chart are frequently revenue, headcount, results and other things that have no business being typed into a stranger's website.

Related tools

Color Palette Generator generates a coherent colour set for a multi-series chart. Resize Image adjusts the finished image to exact dimensions. Add Text to Image adds a caption or annotation on top of the downloaded chart.