Sorting a bar chart by value hands readers the ranking and the extremes for free

What it is

Sort a bar chart by value and two pieces of information appear free: the rank order (who beats whom) becomes the position order, and the maximum and minimum land at the ends. Readers compare no lengths at all — "who is first, who is last, and how wide is the gap" is answered by position alone. This is the highest-return readability upgrade available in a single chart, usually one line of sorting code.

Why it happens

Value sorting works by giving one variable a second encoding channel: bar length encodes the value, and the post-sort position redundantly encodes that same value's order. The visual system locates extremes almost preattentively — the tallest and shortest bars sit at the field's endpoints, no bar-by-bar comparison needed. An unsorted chart forfeits both channels: size requires individual length comparisons, position contributes nothing, and "find the maximum" turns from one glance into one scan. Sorting also organises the middle tiers into legible shapes (a steep drop, a plateau, a long tail) that are invisible in scrambled order.

Where it stops holding

Value sorting only makes sense for nominal categories; re-sorting ordered categories by value is destruction (next card). Sorting also has a price: once ordered by value, "look up a specific category by name" gets harder, and readers accustomed to alphabetical lookup need interactive search or a supporting index. When both jobs matter — ranking and lookup — one order cannot optimise both; sort for the primary task and hand the secondary one to interaction.

Applying it

• Default bar charts to value order (largest at top) unless categories carry intrinsic order or the task is a named lookup.
• After sorting, check the ends: maximum and minimum should sit exactly at the chart's visual endpoints.
• Verification: ask readers to name the largest and smallest categories without pausing; an instant answer means the sort holds — visible scanning means the sort is missing or broken.

Related

• Same group: U5.04.1 Category-axis order is a designer's choice, not a property of the data · U5.04.3 Categories with intrinsic order must not be re-sorted by value · U5.04.4 Side-by-side charts need a shared order for cross-chart comparison · U5.04.5 State the sorting basis in the axis title or caption
• Nearby: U1.01.1 Visual channels show a stable empirical accuracy ordering · U5.02.4 Too many legend entries degrade matching into a linear search
• Search terms: sort bars by value · rank readability · preattentive endpoints