4.1.1. Selecting, Formatting, and Configuring Visuals
💡 First Principle: Each visual type encodes data in a way human perception reads well for one kind of question. Position along a common axis (bars, lines) is read most accurately; angles and areas (pies, donuts, treemaps) are read least accurately. Matching visual to question is matching encoding to perception.
| Business question | Good visual choices | Avoid |
|---|---|---|
| Compare categories | Clustered bar or column chart | Pie with many slices |
| Trend over time | Line chart, area chart, column chart for few periods | Bar chart with dates on the vertical axis |
| Part-to-whole | Stacked bar/column, 100% stacked, donut or pie (few categories), treemap | Pie with more than about 5 slices |
| Relationship between two measures | Scatter chart (bubble size for a third) | Two separate bar charts |
| Single key number | Card, multi-row card, KPI (value vs. target with trend) | Table |
| Progress to target | Gauge, KPI, bullet-style bar with target line | Pie |
| Detailed values, cross-tab | Table, matrix (with drill and subtotals) | Charts with hundreds of labels |
| Geographic distribution | Map, filled map, shape map, Azure Maps | Bar chart when location matters |
| Contribution to change | Waterfall chart | Stacked column |
| Stages of a process | Funnel | Line |
| Two measures with different scales | Line and clustered column combo chart | Two overlapping charts |
Formatting happens in the Format pane (Format visual and General tabs): titles and subtitles, axis ranges and labels, data labels, legends, gridlines, colors per series or category, borders, shadows, headers, and visual tooltips. Configure, don't decorate: remove clutter (redundant axis titles, heavy gridlines), label directly where possible, and use consistent colors for the same category across visuals.
Configuration choices the exam names include small multiples (repeat a chart per category in a grid), drill hierarchies on axes (drill down, expand all, the drill-mode buttons in the visual header), Show items with no data (display categories that have no matching facts), Top N filters on a visual, a reference or target shown with a constant line or a KPI's target well, and field parameters, which let readers switch the measure or dimension a visual shows from a slicer.
For tables and matrices, know stepped layout, subtotals and grand totals on/off, Show on rows for multiple measures, and expand/collapse row headers.
⚠️ Exam Trap: For comparing values across many categories, the expected answer is a bar or column chart, not a pie or donut. Pies are acceptable only for part-to-whole with very few categories. When a stem says "compare sales across 15 product categories," eliminate pie, donut, and treemap first.
Reflection Question: A manager wants to see monthly revenue alongside profit margin percentage on one visual. Which visual handles two measures on different scales, and why would a clustered column chart with both measures mislead?
