Design Recommendations for Chart Types
Area Chart | Column/Bar Chart | Segmented
Column/Bar Chart | Line Chart | Pie
Chart | Scatter Plot
Area Chart

Figure 1: Area chart
Types
- Percentage: The sum always represents 100% (relative scale)
- Cumulative: The sum can vary according to the elements (absolute scale)
Layout Recommendations
- Position the largest area lowest
- Position darkest color or pattern lowest (use lighter colors when moving
up)
- Emphasize data through colors (redundant coding)
Column/Bar Chart

Figure 2: Multiple column chart
Layout Recommendations
- The baseline of a bar chart or column chart consists of labels (e.g. customers,
regions, products), not a numerical scale, even if it sometimes looks like
that.
- Types:
- Column Chart: Vertical columns
- Use if the values are more important than the labels (e.g. years)
- Bar chart: Horizontal bars
- Use if the labels for the data are very long
- Use if the information is more important than the values themselves
-> the labels are to the left and are read first
- Column/bar charts can be used to compare more than one category at a time
-> multiple column/bar charts.
- Separate columns/bars through white space (for more than one categories,
only the groups are separated through white space)
- Choose an organizational principle for the baseline:
- Time, e.g. costs from January to December
- Quantity or volume, e.g. number of sold units
- Alphabet
- None: order columns/bars according to rank in ascending or descending
order
- If possible show comparison values like a "baseline" value or a "critical" value
- If applicable and possible show error limits
Recommendations for Labels
- Add labels to axes and columns/bars: Place scale and marks at the axis,
label below (columns) or left to (bars) the baseline.
- For multiple column charts:
- Place group labels below the baseline, left-aligned to the columns
- Column labels:
- Place label for each column type into the diagram area or
- Use shading or patterns plus a key to identify the columns
- Label multiple bar charts analogously
- Bar charts
- Short labels: Place labels to the left
- Long labels: Right-align labels to the right chart border
Look
- Use a unique color or pattern for each value; use a contrast of at least
30% between neighboring columns or bars
- Striped patterns: Use patterns with an angle of 45° to the base line
in order to avoid optical illusions; let the stripes all go into the same
direction
Better: Avoid striped patterns and use solid colors or grays
- If possible use colors or patterns that emphasize the data with similar
meaning, e.g. red for "critical" or "red shirt"
Segmented Column/Bar Chart

Figure 3: Segmented column chart (relative values)
Recommendations
- For scales, axes and layout see column/bar charts
- Arrange segments within each column/bar in the same order
- Do not place labels into the segments. Place them right beside the diagram
or above it - or use a legend. Do not label each column/bar
- If possible, place the segment with the least change lowest and the most
variable at the top; for bars do the analog. This arrangement makes comparison
easier and emphasizes changes
- Make the segments easy to distinguish; use the most striking color or pattern
for the most important data
- Limit the number of segments to those that are large enough to be easily
recognized and labeled; combine small segments into a larger one
Line Chart

Figure 4: Line chart
Layout Recommendations
- Use the Y axis for variable (dependent), the X axis for fixed (independent)
information
- For comparisons of data sets, draw each data set as a graph of its own
in a common line chart (not more than 4-5 graphs - otherwise split data sets
into several charts:
- For example, compare the most important graph with every other graph
- Plot the most important graph against a small set of other relevant
graphs
- There are no rules for the interval size of the axes; however, the graph
should be neither too steep nor too flat
- Label each graph, if possible (otherwise use keys)
- If appropriate, display a reference line
- For multiple graphs: Use symbols for the data points to visually set the
graphs apart; make the symbol 2-3 times larger then the line
Look
- If there are several graphs within one chart, use colors or line styles
to distinguish the lines (care for overlapping graphs)
- Emphasize the most relevant graph, if applicable, by using the most striking
color or the thickest lines for it
- If lines cross each other, interrupt the background lines
Special Variants
- Graph with double-logarithmic or half-logarithmic scale divisions
- Graph with variance bars, stock charts (High/Low/Close) etc.
Pie Chart

Figure 5: Pie chart
Layout Recommendations
- Start at 12 o clock (top), then follow the direction of the clock
- The values must add up to 100% (or have to be scaled accordingly)
- Do not use too small circle segments (> 18° or 5%); combine small
segments intro larger ones, if applicable
- Label each segment
- Labels can be placed within segments (if these are large enough and the
labels short) or outside (small segments, long labels)
- Write the values beside the segments, because angles are hard to estimate
Look
- Use highlighting methods only, if really necessary. Typical emphasizings
are:
- Pull a segment out of the pie
- Use different colors or patterns
- Patterns: Try to alternate between patterns and solid colors; avoid vibrating
patterns or patterns that produce optical illusions.
- Use perspective only, if distortions are tolerable (write values beside
the segments in this case).
Scatter Plot
Figure 6: One-dimensional scatterplot (left), two-dimensional
scatterplot (right)
Look
- Preferably use a square for the diagram; for two-dimensional scatter plots
both axes should have identical scale divisions
- Use different symbols, if different sets of data are included in one scatter
plot:
- Useful symbols are:
;
Less useful symbols are:

- Different colors help to distinguish symbols; circles are good distinguishable
from triangles, but not good distinguishable - especially if small -
from squares
- Colors may be used "symbolically", e.g.
= "without
treatment",
= "with
treatment"
- All symbols should be of the same size - or at least look as if they
were of same size (sometimes optical corrections are necessary)
- Use overlapping symbols for overlapping data points, not just one symbol
- Identical data points should be represented through special symbols
which indicate this fact
- Important or critical data points should be highlighted, e.g. though more
intensive color, a different color or a different shape
- If data points lie on the axes, move the axes slightly outwards
top