Chart - Box Plot Distribution.

Best for: Comparing variability in sales cycle length, diagnosing swings in response or handling time, reporting quality spread across teams, branches, or machines, proving the variation an average hides, assessing predictability

Use: When the same metric has been measured across two to four groups and the point is which group is consistent and which is erratic. Use it for sales cycle length, response and handling time, or unit cost, where the averages look alike but the spreads do not. Two preconditions apply. First, the groups must share a single scale and every figure must fall inside scale_min to scale_max, because separate axes per group void the box-width comparison. Second, min < Q1 < median < Q3 < max must hold for each row, and the percentage positions are precomputed by the author against the scale range.

Avoid: Use chart-bar when one representative value per group is enough, chart-line for one metric over time, and chart-slope-comparison for per-item change between two points in time. Use chart-scatter-quadrant to scatter individual items across two axes, chart-heatmap for the intensity at a row-by-column crossing, and chart-mix-shift-columns for a shift in composition. A single group has nothing to compare against, and past five groups the rows compress until the quartile figures stop reading. Do not use it for a metric with too few observations to support quartiles.