Which Charts Work for a Small Number of Public Metrics?
From the earlier reference collection. Read the current software blog.
Choose a chart around the question—comparison, change over time, or distribution—and keep small datasets easy to read.
Use bars to compare categories
When the question is which of a few places or groups has more, a bar chart is often direct. Start the axis at zero when bar length represents magnitude, sort categories in a meaningful order, and label the measure and unit. If there are only three values, a clear table may communicate them just as well.
For example, a dashboard comparing reported incidents in four districts can use a horizontal bar chart with counts at the ends. If the districts have different populations, raw counts may not answer a fairness or risk question; a rate with a stated denominator may be more appropriate.
Use a line for ordered time
A line chart fits values that change across a sequence, such as hourly readings. Keep time intervals clear and show gaps where observations are missing. Connecting across a long outage can falsely imply that the value remained stable. If observations are irregular, note that rather than presenting evenly spaced points as uniform sampling.
With only a few time points, direct labels or a table can be easier to interpret than a decorative chart. The horizontal axis should identify the time zone and interval, especially when a day crosses midnight or readers may compare locations.
Use distributions to show spread
A single average can hide a wide range of readings. When the spread matters, show individual points, a range, or a simple distribution view. For a handful of sites, a dot plot with one labeled point per site can preserve detail better than a complex box plot whose summary statistics readers may not need.
A map is helpful when location is central to the question, but it should not be used as a substitute for a comparable scale. A map with differently sized symbols can exaggerate differences if the size rule is not clear. Pair geographic display with a table or legend that states units and values.
Prefer legibility to chart variety
A small dashboard does not need a different chart type for every metric. Reusing a consistent visual language makes it easier to scan. Check that color is not the only way to distinguish series, that labels remain readable on narrow screens, and that the same scale is used when panels invite direct comparison.
Choose the simplest view that answers the reader’s question. If the chart needs a long explanation to decode, a labeled table may be the better instrument.