Tables are the oldest form of data display and the most dismissed. In a world obsessed with interactive dashboards and animated transitions, the table sits in the corner, unsexy and indispensable.
Tufte defended tables more vigorously than most of his readers realize. In The Visual Display of Quantitative Information, he argued that for small datasets — twenty numbers or fewer — a well-designed table almost always outperforms a chart. The reader can see exact values, make precise comparisons, and scan rows and columns in any order. No chart offers all three.
When Tables Win
Tables win when the reader needs exact values. A bar chart shows that Product A's revenue is "around $3.4 million." A table shows it is $3,412,000. For financial reporting, scientific data, and operational metrics, "around" is not good enough.
Tables win when multiple attributes must be compared simultaneously. A chart can show revenue by product. A table can show revenue, margin, growth rate, and market share by product, all in one glance. The chart would require four separate panels. The table needs four columns.
| Product | Revenue | Margin | Growth |
|---|---|---|---|
| Alpha | $3.4M | 42% | +12% |
| Beta | $2.1M | 38% | +3% |
| Gamma | $1.8M | 51% | -7% |
| Delta | $0.9M | 29% | +22% |
Four products. Four attributes. Sixteen numbers. A reader can scan this table in seconds and form a complete picture. No chart could match this information density at this level of precision.
Tables also win when the audience will use the display as a reference. Nobody looks up a specific value in a bar chart. People look up values in tables constantly. Financial statements, sports statistics, timetables, price lists — these are tables because tables are the right tool.
Designing Good Tables
Most tables are poorly designed. They use heavy borders, inconsistent alignment, too many colors, and fonts that fight the numbers. A well-designed table is a quiet, efficient object.
Right-align numbers. This allows the reader to compare magnitudes by scanning the column. Left-aligned numbers force the reader to find the decimal point in each row independently.
Use light horizontal rules, no vertical rules. Vertical rules add clutter without aiding comprehension. Horizontal rules separate rows and guide the eye across columns. This is one of Tufte's specific recommendations, and it is correct.
Minimize the number of decimal places to what the data supports. Reporting revenue to the penny in a table of millions is noise. Round to the appropriate level — thousands for small companies, millions for large ones.
Sort meaningfully. The default sort — alphabetical by name — is almost never the most useful. Sort by the variable the reader cares about: highest revenue first, fastest growth first, largest decline first. The sort order is an editorial choice that shapes how the reader encounters the data.
The Hybrid: Sparklines in Tables
Tufte proposed embedding small word-sized charts — sparklines — directly in table cells. A table of monthly revenue by product gains enormously from a tiny line chart in a "Trend" column. The reader gets both the exact current value and the historical trajectory in a single row.
This hybrid form combines the precision of tables with the pattern-recognition advantage of charts. It remains underused, partly because few tools make it easy to create and partly because few designers think to try.
The Case Against Chart-Everything
The instinct to visualize every dataset is misguided. Some data is better served by a well-formatted table than by any chart. The question to ask is not "what chart should we use?" but "does this data need a chart at all?"
When the dataset is small, when exact values matter, when multiple attributes must be compared, and when the reader needs a reference tool — the answer is often no. The answer is often a table.