What is data visualization? Chart types, principles and examples

What is data visualization?

Data visualization is the practice of representing data graphically, as charts, graphs, maps and dashboards, so that people can see patterns, trends, comparisons and outliers quickly and accurately. A good visualization answers one question at a glance; a table of the same numbers makes the reader do the work.

It sits between analysis and communication. Analysts use it to explore ("is there a weekly pattern?"), and teams use it to explain ("signups dropped after the pricing change").

Why it matters

Summary statistics hide shape. Anscombe's quartet is four small datasets with the same mean, variance and correlation that look nothing alike when plotted: one is a clean line, one is a curve, one has a single outlier, one is a vertical stack with one extreme point. Plotting the data first is how you notice that an average is misleading.

Which chart answers which question

How people read charts

Cleveland and McGill (1984) tested how accurately people judge values from different visual encodings. From most to least accurate:

1. Position along a common scale (bars, dots, lines on one axis)
2. Position along non-aligned scales (small multiples)
3. Length, direction, angle
4. Area
5. Volume, curvature
6. Colour saturation and hue

The practical rules follow directly: prefer bars and lines to pies, bubbles and 3D; use colour to highlight, not to encode precise values; when comparing, put things on the same axis.

A worked example: pie vs sorted bar

Sample data: 12,000 sessions in a month by channel.

As a seven-slice pie, Referral (15%) and Paid search (12%) look about the same, and Email and AI sources are slivers needing a legend lookup. As a horizontal bar chart sorted by sessions with values labelled at the end of each bar, the ranking, the gaps and the long tail are clear without a legend. Same data, but the second chart can be read correctly in a few seconds.

Best practices

• One chart, one message. Write the takeaway as the title: "Organic search brings a third of sessions", not "Sessions by channel".
• Start bar axes at zero. Line charts may zoom in, bars may not.
• Label directly instead of using a legend where possible.
• Limit colours. Grey for context, one accent colour for what matters.
• Show context: the previous period, a target or a benchmark.
• Annotate events such as releases and campaigns on time series.
• Avoid dual-axis charts unless both series truly share a story; they invite false correlations.

Common mistakes

• Truncated bar axes that turn a 3% difference into a visual doubling.
• 3D effects, which distort length and area.
• Rainbow palettes on ordered data.
• Too many series on one line chart ("spaghetti").
• Cumulative charts used to hide a recent decline.
• Percent changes on tiny bases ("+200%" from 1 to 3).

Tools

Spreadsheets (Excel, Google Sheets) cover most everyday charts. BI tools (Looker Studio, Power BI, Tableau) add interactivity and sharing. Code libraries (matplotlib, ggplot2, D3) give full control. Analytics products have built-in charts for their own data, which is usually the fastest route for traffic and product questions.

Data visualization in TapCub

Reports in TapCub are charts with tables underneath, and the tools above apply to them: compare date ranges to show context, add chart annotations for releases and campaigns, then share a dashboard or schedule email reports. For turning charts into a narrative, see data storytelling.

Sources

• Cleveland, W. S. and McGill, R. (1984). Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods. Journal of the American Statistical Association, 79(387), 531–554.
• Anscombe, F. J. (1973). Graphs in Statistical Analysis. The American Statistician, 27(1), 17–21.
• Tufte, E. R. (1983). The Visual Display of Quantitative Information. Graphics Press.