Perception

Perception isn’t neutral: people actively build what they see. Gestalt psychology describes how people group visual elements into wholes, looking for order and simplicity. Each principle below comes with what it means for design.

Gestalt principles

Proximity: space signals structure

People group elements that are close together. Even small differences in spacing create strong perceived structure.

• Group related rows or columns in a table with tighter spacing, and separate groups with white space instead of lines.
• Cluster related KPIs on a dashboard.

Example: in a table of monthly figures, tighter spacing within quarters and more space between them shows the quarters without any borders.

Similarity: shared attributes bind meaning

Objects with the same colour, shape, size or orientation look related.

• Use one highlight colour consistently for the “story” data.
• Colour is a strong channel that is easily overloaded, so don’t overuse it.

Example: to feature one region out of twelve, make all bars grey and the featured region Teal.

Enclosure: boundaries create groups

People treat everything inside a boundary as a group.

• Light shading can separate forecast from actual values.
• Use enclosure strategically, not as a box around every chart.

Closure: readers complete shapes themselves

People complete shapes mentally, so frames and borders are unnecessary.

• Remove chart borders, heavy gridlines and background shading. A bar chart without a frame still reads as a whole.

Continuity: the eye follows smooth paths

The eye follows the smoothest path, and sees continuity even where there is none.

• Align labels so the eye moves naturally.
• Drop the y-axis line: aligned tick labels or gridlines already imply it.

Connection: the strongest grouping cue

Connected elements look like they belong together, and connection beats similarity.

• Use lines deliberately, as in a time series, and don’t connect things that don’t belong together.

Figure and ground: contrast is relative

Perception depends on relative brightness and local contrast.

• Give the focal data the strongest contrast, and put secondary elements on soft backgrounds.
• A bright highlight on a busy background loses its effect.

Pre-attentive attributes

People spot some features almost instantly, before conscious attention: colour, size, orientation and shape.

• Use a single pre-attentive feature (colour or size) to highlight one insight.
• Combine channels only when the data structure really calls for it, which is rare.
• Shape stands out less than colour, especially among many points.

Example: show an outlier in a scatter plot in one vivid colour, with all other points neutral.

What this means in practice

Rule in the guide	Principle
No borders, light gridlines	Closure
Group with space, not lines	Proximity
One highlight colour, grey for the rest	Similarity, pre-attentive colour
Shade forecasts or periods	Enclosure
Align labels, drop axis lines	Continuity
Connect only related points	Connection
Annotate key insights	Prevents readers from seeing patterns that aren’t there

---

Sources: Healy (2026), Data Visualization: A Practical Introduction; Nussbaumer Knaflic (2015), Storytelling with Data.