Table Design

Laying out rows and columns so the important number is found before everything else.

Also known as: table layout, designing tables, table formatting

A table is a chart with a grid. It is the right form when the reader needs exact values, needs to look up one row, or needs to compare a measure across many rows. Tables get designed badly more often than charts, because they are the default rather than a choice.

What makes one work

The important number is found first. Sort by it by default, keep it in a consistent column, and right-align it. Left-aligned numbers cannot be compared down a column.
Consistent units and consistent decimals. A column mixing “1.2k” with “842” and “12,400” cannot be scanned at all.
One idea per column. If one column holds both this month’s figure and last month’s, the header tells the reader nothing about what to compare.
Fewer columns than you think. Every extra column costs the whole row. Ten columns is a data dump; five is a finding.
Group rows and put a subtotal where each group starts. See pivot table for the spreadsheet way of building one.
Keep the totals row visually distinct from the last group, and label it. A reader should never mistake a total for a member.
Say when the data was refreshed and where it came from — a stale table presented as current is worse than no table, so link or state data freshness.

The trade-off

A table is precise and slow. When the question is “which is biggest” rather than “what is the exact number”, a bar chart answers faster. When the question is “which is moving”, a sparkline column answers it without a second view.

Many tables earn their place by carrying both: the number for the reader who needs it and the shape for the reader scanning. Ordering the columns is visual hierarchy applied to a grid, and the same rule holds — decide what should read first, then let size, weight and spacing express that.